**6. Bilateral internal mammary artery conduits**

In the 1980s, the LIMA-to-LAD graft was shown to be an independent predictor of improved short- and long-term results when used as a conduit for CABG compared to RSVG-only grafting [76]. Unequivocal advantages of the LIMA-LAD graft include prolonged survival relative to use of RSVG to LAD, reduced rates of recurrent angina, reduced postoperative MI and other ischemic events, and decreased need for coronary reintervention [76, 84]. The superiority of the LIMA in comparison to other CABG conduits

[15, 17]. Similarly, Guru et al evaluated the potential benefit of multiple arterial grafting in over 53,000 patients undergoing primary CABG between 1991 and 2001. After propensity matching, patients receiving 2 arterial grafts had decreased rates of cardiac readmission and reduced incidence of the composite of cardiac readmission, death, and repeat revascularization relative to those with one arterial grafts [23]. Furthermore, patients receiving 2 arterial grafts had improved survival compared with patients receiving only one arterial graft [23]. Similar findings were reported by Nasso, et al, who found no differences at 2 years between groups receiving RA, in-situ RIMA, or free RIMA as the 2nd arterial graft, although each of these groups was superior to patients receiving only one arterial graft (LIMA-LAD) with respect to cardiac event-free survival [77]. Zacharias et al also demonstrated advantages of RA grafting as a 2nd arterial conduit on long-term survival

In addition, multiple arterial grafts and their arrangements in all coronary distributions have been proven superior to venous grafts with regard to long term patency regardless of the anatomic details of the native coronary and distal anastomosis [55]. These results are particularly applicable in the context of recurrent angina [16, 55, 78]. Finally, perhaps one of the best recent demonstrations of the advantages of arterial grafting over RSVG conduits was provided by Gaudino et al, who studied 60 CAD patients who had previously undergone PCI and developed in-stent restenosis. After undergoing CABG, patients receiving IMA and RA grafts had patency rates of 90% while those undergoing RSVG had

Since SVG conduits inevitably fail, particularly late [62], there has been increased enthusiasm for total arterial revascularization for CABG. Total arterial revascularization may obviate the concerns of vein graft failure and has been shown to have good shortterm results [80]. However, little evidence is available to suggest that outcomes are improved with "all-arterial" grafting [81]. Zacharias et al have recently demonstrated that patients with multi-vessel CAD undergoing all-arterial grafting had improved 12-year survival compared with matched patients who underwent standard CABG with LIMA-LAD and RSVG to other distal targets [82]. Furthermore, complete coronary revascularization and use of all-arterial grafting strategy was associated with improved

It has been estimated that all-arterial grafting is possible in 90% of patients using various conduits and their configurations [55], and even patients with advanced age have been shown to benefit from all-arterial revascularization strategies in terms of freedom from

In the 1980s, the LIMA-to-LAD graft was shown to be an independent predictor of improved short- and long-term results when used as a conduit for CABG compared to RSVG-only grafting [76]. Unequivocal advantages of the LIMA-LAD graft include prolonged survival relative to use of RSVG to LAD, reduced rates of recurrent angina, reduced postoperative MI and other ischemic events, and decreased need for coronary reintervention [76, 84]. The superiority of the LIMA in comparison to other CABG conduits

when compared to RSVG conduits [14].

**5.1 Total arterial revascularization** 

12-year survival [82, Figure].

patency rates of 50% at a mean follow-up of 52 months [79].

recurrent coronary events and improved graft patency [83].

**6. Bilateral internal mammary artery conduits** 

Fig. 1. Effects of completeness of revascularization on 12-year old Kaplan-Meier survival in triple-vessel disease (3-Ves Dis) patients. (Left) All-arterial patients. (Right) Internal thoracic artery/saphenous vein (ITA/Vein) patients. Incomplete - completeness of revascularization index (CRI) less than 1, or 2 grafts; complete CRI equal to 1, or 3 grafts; complete plus - CRI greater than 1, or 4 or more grafts. All p values by log-rannk (Mantel-Cox) test. (CABG coronary artery bypass graft surgery.)

may be related to its unique freedom from arteriosclerosis and due to the rich run-off bed provided by the LAD coronary and its branches [85]. Since there is no basis for suggesting or concluding that the biological and mechanical properties of the right IMA are different from the LIMA, successes with the LIMA have prompted investigation of the potential benefits of bilateral IMA (BIMA) grafting.

The original description of BIMA for CABG is credited to Kay in 1969 [86]. Since then, multiple centers including our own have investigated the impact of BIMA grafting on longterm results of CABG. Advantages of BIMA have been somewhat difficult to prove definitively without randomized controlled trials in this area, which have not been conducted secondary to cost concerns and administrative requirements associated with studies inherently requiring significant longitudinal follow-up [87]. Instead, investigation and documentation of BIMA benefits have relied on evaluating institutionally maintained observational databases to show differences between the "treatment group" and the "control group" by way of propensity matching [87]. Analysis of these data show improved long-term results for patients receiving BIMA grafting as compared with single IMA grafting. However, survival curves do not separate until several years postoperatively, which has been a consistent finding [15, 88, 89; Figure]. The demonstrated clinical advantages of BIMA grafting strategies include prolonged survival and reduced need for coronary reintervention on the basis of recurrent myocardial ischemia, including freedom from the need for coronary reintervention [15, 88, 90] which hold true for women as well as for men, where it has been demonstrated that use of BIMA had 3-fold improved cardiacrelated survival compared with patients who did not receive an IMA graft [91].

Reported rates of BIMA use in CABG range from 4.0% to nearly 50% depending upon several factors including the contributing authors' practice preferences and the particular patient cohort treated [19-22, 92]. However, it has been estimated that up to 80% would be candidates for BIMA grafting [93]. Subjective and potential obstacles to BIMA use include increased surgical times, increased technical challenges, especially related to the positioning

Conduit Selection for Improved Outcomes in Coronary Artery Bypass Surgery 151

point out that if strict criteria for selecting patients for BIMA grafting had been used, nearly

As noted previously, RSVG conduits for CABG have not been compared directly to DES, but comparisons of BIMA grafting with a strategy primarily employing DES for coronary revascularization do exist [102, 103]. After matching for multiple comorbidities, significantly improved angina, reduced need for reintervention, and improved reintervention-free survival at one year have been observed among patients undergoing BIMA CABG relative to PCI [102]. In a similar study, Locker et al examined BIMA CABG versus PCI in diabetic patients and demonstrated more complete coronary revascularization, improved angina, reduced need for reintervention, and increased cardiovascular event-free survival (80% vs. 30%, p <0.001) in patients undergoing BIMA CABG [103]. More importantly, 6-year survival among patients with left main CAD or 3-vessel CAD was significantly better with BIMA revascularization, providing rare evidence for superiority of multiple arterial grafting

One controversial and often-debated point concerning BIMA CABG relates to the native coronary artery planned as a target for the second (right) IMA. Initial opinion considered best results of BIMA CABG to occur when the RIMA was grafted to the "next most" important coronary bed angiographically, assuming that LIMA to LAD was nearly always performed [104]. Schmidt et al found that multiple left-sided grafts lead to improved survival relative to a cohort of patients in which the RIMA was placed to the right coronary artery distribution along with LIMA-LAD grafting [104]. However, it should be noted that graft patency was quite good in both groups (91.7 % vs. 89.6%), as was survival, and freedom from heart failure, angina and need for reintervention [104]. They argued that maximum long-term benefit from BIMA grafting is realized when the 2nd IMA is placed to the coronary distribution subtending the most amount of viable myocardium, which was

The requirement for the 2nd IMA targeted to the left coronary system has been refuted by other data. Analysis of Cleveland Clinic data shows that benefits of BIMA CABG are similar regardless of the recipient coronary artery for the 2nd IMA graft [89]. Sabik et al found that risk-adjusted and unadjusted outcomes did not differ between a strategy for the 2nd IMA going to the right coronary or the circumflex coronary artery territory, but several important caveats are applicable, particularly with regard to the right coronary artery if the RIMA is to be used as an in situ graft to this coronary distribution. For instance, the distal RCA should be free of disease, and the proximal RCA stenosis should be 70% in diameter to avoid competitive native coronary flow, which may be detrimental to IMA patency. Finally, distal viable myocardium should be ensured [105]. More recently, Kurlansky et al reviewed their experience in over 2,200 consecutive patients who underwent BIMA grafting. In 2/3 of patients, the 2nd IMA was placed to the left coronary system, while in 1/3 of patients, the 2nd IMA was placed to the right coronary system. Incredibly, 98% of patients had complete coronary revascularization performed with in-situ arterial configurations, that is, avoiding free IMA utilization. At a mean follow-up of nearly 13 years, long-term survival was no

Gansera et al recently described their strategy for with BIMA versus single IMA in which the RIMA was usually directed anteriorly and to the left as pedicled conduit for the LAD, while the LIMA was typically directed for revascularization of the lateral wall (circumflex artery distribution) [95]. Importantly, RIMA crossover grafts to the LAD coronary system appear to have patency rates that are not different from LIMA-LAD configurations [107]. In

70% of patients in their series would have been excluded [24].

strategy relative to PCI [103].

most commonly the circumflex artery distribution.

different between groups [106, Figure].

#### Fig. 2.

of the non-LIMA-to-LAD graft, and increased rates of sternal wound complications relative to patients with one or less IMAs harvested [94]. After 10 years of experience with BIMA CABG, Gansera et al noted increased OR and aortic cross clamp times among BIMA patients compared with single IMA CABG patients. In addition, patients receiving BIMA grafting had higher rates of bleeding requiring postoperative mediastinal reexploration (2.9% vs. 0.6%) along with increased rates of wound complications [95]. However, they also noted that nearly one full additional distal graft was completed when both IMAs were used and, most importantly, BIMA grafting was associated with improved 30-day survival, particularly among diabetic patients, compared with only one IMA graft. [95]. In this same study, a grafting strategy not incorporating 2 IMA conduits was an independent predictor of *perioperative* mortality, directly disputing biases that BIMA grafting is associated with increased perioperative complications [95]. Similarly, Kurlansky et al reviewed their collective experience with more than 4,500 consecutive CABG procedures (2,369 single IMA; 2,215 BIMA) and demonstrated that hospital mortality was significantly reduced among patients undergoing BIMA grafting compared with controls [92].

Nevertheless, selection of patients for BIMA grafting should be performed cautiously, particularly when performing surgery through a standard median sternotomy incision. This is primarily related to the well-demonstrated risk for sternal wound infection after BIMA grafting in patients with diabetes, obesity, and other comorbidities [96-100]. However, BIMA grafting can be performed safely without significant increased risk for sternal wound infection as has been frequently demonstrated [24, 98, 100, 101]. For example, Jones et al reviewed their experience with 500 consecutive patients undergoing BIMA CABG over an 11-year period, excluding only those with proximal coronary stenoses <70% and emergency cases with HD instability. Nevertheless, the reported rates of perioperative complications were low: operative mortality was 1.8%; deep sternal wound infections were 1%; and 1.8% had take back for bleeding. Incredibly, only 2 patients out of the 500 required reoperation for myocardial ischemia in the follow-up period. [24]. Interestingly, however, Jones et al

of the non-LIMA-to-LAD graft, and increased rates of sternal wound complications relative to patients with one or less IMAs harvested [94]. After 10 years of experience with BIMA CABG, Gansera et al noted increased OR and aortic cross clamp times among BIMA patients compared with single IMA CABG patients. In addition, patients receiving BIMA grafting had higher rates of bleeding requiring postoperative mediastinal reexploration (2.9% vs. 0.6%) along with increased rates of wound complications [95]. However, they also noted that nearly one full additional distal graft was completed when both IMAs were used and, most importantly, BIMA grafting was associated with improved 30-day survival, particularly among diabetic patients, compared with only one IMA graft. [95]. In this same study, a grafting strategy not incorporating 2 IMA conduits was an independent predictor of *perioperative* mortality, directly disputing biases that BIMA grafting is associated with increased perioperative complications [95]. Similarly, Kurlansky et al reviewed their collective experience with more than 4,500 consecutive CABG procedures (2,369 single IMA; 2,215 BIMA) and demonstrated that hospital mortality was significantly reduced among

Nevertheless, selection of patients for BIMA grafting should be performed cautiously, particularly when performing surgery through a standard median sternotomy incision. This is primarily related to the well-demonstrated risk for sternal wound infection after BIMA grafting in patients with diabetes, obesity, and other comorbidities [96-100]. However, BIMA grafting can be performed safely without significant increased risk for sternal wound infection as has been frequently demonstrated [24, 98, 100, 101]. For example, Jones et al reviewed their experience with 500 consecutive patients undergoing BIMA CABG over an 11-year period, excluding only those with proximal coronary stenoses <70% and emergency cases with HD instability. Nevertheless, the reported rates of perioperative complications were low: operative mortality was 1.8%; deep sternal wound infections were 1%; and 1.8% had take back for bleeding. Incredibly, only 2 patients out of the 500 required reoperation for myocardial ischemia in the follow-up period. [24]. Interestingly, however, Jones et al

patients undergoing BIMA grafting compared with controls [92].

Fig. 2.

point out that if strict criteria for selecting patients for BIMA grafting had been used, nearly 70% of patients in their series would have been excluded [24].

As noted previously, RSVG conduits for CABG have not been compared directly to DES, but comparisons of BIMA grafting with a strategy primarily employing DES for coronary revascularization do exist [102, 103]. After matching for multiple comorbidities, significantly improved angina, reduced need for reintervention, and improved reintervention-free survival at one year have been observed among patients undergoing BIMA CABG relative to PCI [102]. In a similar study, Locker et al examined BIMA CABG versus PCI in diabetic patients and demonstrated more complete coronary revascularization, improved angina, reduced need for reintervention, and increased cardiovascular event-free survival (80% vs. 30%, p <0.001) in patients undergoing BIMA CABG [103]. More importantly, 6-year survival among patients with left main CAD or 3-vessel CAD was significantly better with BIMA revascularization, providing rare evidence for superiority of multiple arterial grafting strategy relative to PCI [103].

One controversial and often-debated point concerning BIMA CABG relates to the native coronary artery planned as a target for the second (right) IMA. Initial opinion considered best results of BIMA CABG to occur when the RIMA was grafted to the "next most" important coronary bed angiographically, assuming that LIMA to LAD was nearly always performed [104]. Schmidt et al found that multiple left-sided grafts lead to improved survival relative to a cohort of patients in which the RIMA was placed to the right coronary artery distribution along with LIMA-LAD grafting [104]. However, it should be noted that graft patency was quite good in both groups (91.7 % vs. 89.6%), as was survival, and freedom from heart failure, angina and need for reintervention [104]. They argued that maximum long-term benefit from BIMA grafting is realized when the 2nd IMA is placed to the coronary distribution subtending the most amount of viable myocardium, which was most commonly the circumflex artery distribution.

The requirement for the 2nd IMA targeted to the left coronary system has been refuted by other data. Analysis of Cleveland Clinic data shows that benefits of BIMA CABG are similar regardless of the recipient coronary artery for the 2nd IMA graft [89]. Sabik et al found that risk-adjusted and unadjusted outcomes did not differ between a strategy for the 2nd IMA going to the right coronary or the circumflex coronary artery territory, but several important caveats are applicable, particularly with regard to the right coronary artery if the RIMA is to be used as an in situ graft to this coronary distribution. For instance, the distal RCA should be free of disease, and the proximal RCA stenosis should be 70% in diameter to avoid competitive native coronary flow, which may be detrimental to IMA patency. Finally, distal viable myocardium should be ensured [105]. More recently, Kurlansky et al reviewed their experience in over 2,200 consecutive patients who underwent BIMA grafting. In 2/3 of patients, the 2nd IMA was placed to the left coronary system, while in 1/3 of patients, the 2nd IMA was placed to the right coronary system. Incredibly, 98% of patients had complete coronary revascularization performed with in-situ arterial configurations, that is, avoiding free IMA utilization. At a mean follow-up of nearly 13 years, long-term survival was no different between groups [106, Figure].

Gansera et al recently described their strategy for with BIMA versus single IMA in which the RIMA was usually directed anteriorly and to the left as pedicled conduit for the LAD, while the LIMA was typically directed for revascularization of the lateral wall (circumflex artery distribution) [95]. Importantly, RIMA crossover grafts to the LAD coronary system appear to have patency rates that are not different from LIMA-LAD configurations [107]. In

Conduit Selection for Improved Outcomes in Coronary Artery Bypass Surgery 153

and others to abandon its use. Subsequently, several RA grafts were empirically observed to be patent at follow-up coronary angiography, leading to the concept's reintroduction in the late 1980s [117]. Since then, the RA has since been widely used and aggressively investigated as a conduit option for CABG due to ease of use, availability in at least 90% of patients, good length allowing reach to any distal target for anastomosis, and it is amenable to concurrent harvesting methods (the IMA, SVG, and RA can be harvested simultaneously) [118]. Depending upon the details of RA grafting, including the target coronary bed and the proximal degree of coronary artery stenosis, long-term RA graft patency approaches 90% and

Since most coronary grafting strategies employing RA grafts do so to evaluate alternatives to SVG conduits, numerous trials comparing RA with SVG for patency and long-term outcomes have been conducted recently [14, 78, 121-123]. In general, these data demonstrate equivalent or improved patency for the RA compared with SVG. One recent randomized controlled trial in the US Veterans Affairs system evaluated RA and SVG as coronary bypass grafts to the "best remaining recipient vessel," with the primary end point of one-year angiographic patency [123]. These data demonstrated equivalent graft patencies for RA compared with SVG, but one-year patency for both groups approached 90% [123]. This is in line with many previous reports on early RA patency, and far exceeds other recent estimations of early graft SVG patency for [8-10]. In contrast, the Radial Artery Versus Saphenous Vein Patency (RSVP) trial compared 5-year graft patency for RA versus SVG when placed to circumflex coronary artery branches that were at least 70% stenotic [121]. These data demonstrated a significantly improved patency for RA grafts (98.3%) relative to

Perhaps the most recognized comparison of RA and SVG is provided by the Radial Artery Patency Study (RAPS), which randomized 561 patients at 13 centers (Canada, New Zealand) to receive a RA graft to the right coronary system or the circumflex coronary system [78]. Saphenous vein grafts were placed to the coronary system not receiving the RA graft, and all patients underwent LIMA-LAD grafting. One-year angiographic follow-up was performed in 440 patients and demonstrated significantly improved patency for RA grafts compared with SVG (91.8% vs. 86.4%, p = 0.009). However, angiographic "string sign" was significantly increased in RA grafts compared with SVG (7% vs. 0.9%, p = 0.001) [78]. It is unclear as to the long-term significance of this finding; as Carpentier originally noted, others have since reported that patency of the RA graft with a "string sign" may actually improve as the native disease worsens [124]. When patient characteristics and target vessel characteristics were considered in the interpretation of the RAPS data, RA grafts were protective of graft patency by multivariable analysis, while smaller distal targets, less proximal coronary artery stenosis, and diabetes were associated with reduced patency rates for SVG or RA grafts [125]. Graft occlusion was significantly more common among diabetics with SVG (19%) compared with RA grafts in diabetics (12%). In fact, RA grafting conferred even greater protection effect from graft occlusion was among diabetic patients than in the

Despite the positive comparisons of RA with SVG patency rates, it has been more difficult to ascribe improved clinical outcomes with RA versus SVG [126]. However, some data suggest that better long-term survival is seen in those receiving RA conduits as compared with SVG (Figure). Zacharias et al evaluated the influence of RA and SVG as a 2nd graft (all patients receiving LIMA-LAD) with regard to survival in 2 groups of 925 matched patients each [14].

can approximate that of pedicled IMA graft patencies [55, 119, 120].

SVG grafts (86.4%) [122].

study cohort as a whole [125].

6-year survival was improved in RA patients [14, Figure.].

Fig. 3. Actuarial survival of optimally matched patients who underwent coronary artery bypass grafting to the left coronary system (LCS [boxes]) and the right coronary system (RCS [triangles]). Number of patients at risk is in parentheses; results are mean = standard error of the mean.

fact, Chow et al reported that RIMA patency was similar to LIMA patency regardless of the coronary distribution to which the artery was grafted [108]. However, a "crossover" configuration for the RIMA places the RIMA-LAD graft at significant risk for injury during subsequent redo sternotomy (as potentially required for aortic valve replacement or other procedures). Consequently, when performing this procedure, we often apply a protective barrier to the mediastinum prior to sternal closure (Repel-CV, SyntheMed, Inc., Iselin, NJ, USA) and often reapproximate the thymic remnant to protect the RIMA-LAD graft [108].

Therefore, the specific BIMA configuration doesn't seem to affect MACE, graft patency, or morbidity/mortality outcomes [110]. Glineur et al evaluated BIMA grafting with both IMA grafts targeted to the left coronary system in either an in-situ or Y-graft configuration, noting equivalent patencies and other mid-term results. In addition, the T-graft configuration has been shown to be safe and effective in grafting multiple distal coronary targets [111].

Most results with BIMA grafting are improved by harvesting the conduit as a "skeletonized" graft rather than as a pedicled graft, in which the accompanying mammary veins and surrounding chest wall muscle and fascia were mobilized along with the artery. Skeletonized IMA harvesting preserves sternal perfusion relative to pedicled IMA harvesting [112, 113]. As a result, skeletonized IMA harvesting is associated with reduced sternal wound complications and longer IMA graft length, allowing more distal coronary targets to be bypassed [Figure, 92, 98, 100]. Skeletonized IMAs also have increased blood flow through the conduit [114], possible as the result of decreased spasm, since skeletonized IMA is typically accomplished without the need for electrocautery on the chest wall [100, 106, 115].

### **7. Radial artery**

Professor Alain Carpentier is credited with introducing the radial artery (RA) as an alternative conduit for CABG [116], but initial results with the RA were unfavorable, leading Carpentier

Fig. 3. Actuarial survival of optimally matched patients who underwent coronary artery bypass grafting to the left coronary system (LCS [boxes]) and the right coronary system (RCS [triangles]). Number of patients at risk is in parentheses; results are mean = standard

fact, Chow et al reported that RIMA patency was similar to LIMA patency regardless of the coronary distribution to which the artery was grafted [108]. However, a "crossover" configuration for the RIMA places the RIMA-LAD graft at significant risk for injury during subsequent redo sternotomy (as potentially required for aortic valve replacement or other procedures). Consequently, when performing this procedure, we often apply a protective barrier to the mediastinum prior to sternal closure (Repel-CV, SyntheMed, Inc., Iselin, NJ, USA) and often reapproximate the thymic remnant to protect the RIMA-LAD graft [108]. Therefore, the specific BIMA configuration doesn't seem to affect MACE, graft patency, or morbidity/mortality outcomes [110]. Glineur et al evaluated BIMA grafting with both IMA grafts targeted to the left coronary system in either an in-situ or Y-graft configuration, noting equivalent patencies and other mid-term results. In addition, the T-graft configuration has

been shown to be safe and effective in grafting multiple distal coronary targets [111].

need for electrocautery on the chest wall [100, 106, 115].

Most results with BIMA grafting are improved by harvesting the conduit as a "skeletonized" graft rather than as a pedicled graft, in which the accompanying mammary veins and surrounding chest wall muscle and fascia were mobilized along with the artery. Skeletonized IMA harvesting preserves sternal perfusion relative to pedicled IMA harvesting [112, 113]. As a result, skeletonized IMA harvesting is associated with reduced sternal wound complications and longer IMA graft length, allowing more distal coronary targets to be bypassed [Figure, 92, 98, 100]. Skeletonized IMAs also have increased blood flow through the conduit [114], possible as the result of decreased spasm, since skeletonized IMA is typically accomplished without the

Professor Alain Carpentier is credited with introducing the radial artery (RA) as an alternative conduit for CABG [116], but initial results with the RA were unfavorable, leading Carpentier

error of the mean.

**7. Radial artery** 

and others to abandon its use. Subsequently, several RA grafts were empirically observed to be patent at follow-up coronary angiography, leading to the concept's reintroduction in the late 1980s [117]. Since then, the RA has since been widely used and aggressively investigated as a conduit option for CABG due to ease of use, availability in at least 90% of patients, good length allowing reach to any distal target for anastomosis, and it is amenable to concurrent harvesting methods (the IMA, SVG, and RA can be harvested simultaneously) [118]. Depending upon the details of RA grafting, including the target coronary bed and the proximal degree of coronary artery stenosis, long-term RA graft patency approaches 90% and can approximate that of pedicled IMA graft patencies [55, 119, 120].

Since most coronary grafting strategies employing RA grafts do so to evaluate alternatives to SVG conduits, numerous trials comparing RA with SVG for patency and long-term outcomes have been conducted recently [14, 78, 121-123]. In general, these data demonstrate equivalent or improved patency for the RA compared with SVG. One recent randomized controlled trial in the US Veterans Affairs system evaluated RA and SVG as coronary bypass grafts to the "best remaining recipient vessel," with the primary end point of one-year angiographic patency [123]. These data demonstrated equivalent graft patencies for RA compared with SVG, but one-year patency for both groups approached 90% [123]. This is in line with many previous reports on early RA patency, and far exceeds other recent estimations of early graft SVG patency for [8-10]. In contrast, the Radial Artery Versus Saphenous Vein Patency (RSVP) trial compared 5-year graft patency for RA versus SVG when placed to circumflex coronary artery branches that were at least 70% stenotic [121]. These data demonstrated a significantly improved patency for RA grafts (98.3%) relative to SVG grafts (86.4%) [122].

Perhaps the most recognized comparison of RA and SVG is provided by the Radial Artery Patency Study (RAPS), which randomized 561 patients at 13 centers (Canada, New Zealand) to receive a RA graft to the right coronary system or the circumflex coronary system [78]. Saphenous vein grafts were placed to the coronary system not receiving the RA graft, and all patients underwent LIMA-LAD grafting. One-year angiographic follow-up was performed in 440 patients and demonstrated significantly improved patency for RA grafts compared with SVG (91.8% vs. 86.4%, p = 0.009). However, angiographic "string sign" was significantly increased in RA grafts compared with SVG (7% vs. 0.9%, p = 0.001) [78]. It is unclear as to the long-term significance of this finding; as Carpentier originally noted, others have since reported that patency of the RA graft with a "string sign" may actually improve as the native disease worsens [124]. When patient characteristics and target vessel characteristics were considered in the interpretation of the RAPS data, RA grafts were protective of graft patency by multivariable analysis, while smaller distal targets, less proximal coronary artery stenosis, and diabetes were associated with reduced patency rates for SVG or RA grafts [125]. Graft occlusion was significantly more common among diabetics with SVG (19%) compared with RA grafts in diabetics (12%). In fact, RA grafting conferred even greater protection effect from graft occlusion was among diabetic patients than in the study cohort as a whole [125].

Despite the positive comparisons of RA with SVG patency rates, it has been more difficult to ascribe improved clinical outcomes with RA versus SVG [126]. However, some data suggest that better long-term survival is seen in those receiving RA conduits as compared with SVG (Figure). Zacharias et al evaluated the influence of RA and SVG as a 2nd graft (all patients receiving LIMA-LAD) with regard to survival in 2 groups of 925 matched patients each [14]. 6-year survival was improved in RA patients [14, Figure.].

Conduit Selection for Improved Outcomes in Coronary Artery Bypass Surgery 155

notoriously susceptible to competitive flow within the native coronary circulation, as may occur when native coronary stenosis is not severe ( 70%) [128, 129]. Based on this frequent observation, use of the RA is not recommended as a conduit for CABG unless the native coronary artery stenosis is high grade ( 70%). This strategy has been associated with improved patency and reduced "string sign" in the RAPS study [78]. Radial artery graft patency is also dependent upon coronary target location, as has been noted for the RIMA graft [89]. For instance, Maniar et al found that grafts to the right coronary artery were more likely to be occluded compared with those placed to the LAD or to the circumflex coronary

Certain inherent, unique characteristics of the RA may also contribute to "string sign" formation [131]. Limb arteries such as the RA are known to be more prone to spasm than somatic (IMA) or splanchnic arteries [132]. One key difference between the IMA and the RA is that the RA is significantly more muscular, leading to increased tendency for spasm requiring prolonged vasodilation [133]. Vasodilation of the harvested RA should begin intraoperatively by exposing the conduit to papaverine or verapamil/nitroglycerin. Verapamil/nitroglycerin may be more effective than papaverine in regards to degree of vasodilation and preservation of endothelial function, which can be a problem with papaverine (especially if injected intraluminally) and with the alpha-blocking agent phenoxybenzamine [118]. Postoperatively, most authors have recommended vasodilation with calcium channel blocking agents or long-acting nitrates for at least one month after

It is debated as to whether the proximal RA graft anastomosis should be performed to the ascending aorta or as a composite graft to other conduits. In a study of over 1,500 radial artery grafts, Maniar et al found no difference between these two types of grafting strategies [129]. Collins et al, reporting the results of the Radial Artery Versus Saphenous Vein Patency (RSVP) trial, demonstrated superior patency of RA grafts compared with SVG, and all proximal anastomoses were performed to the aorta directly [122]. Jung et al recently demonstrated using postoperative CT angiography that RA patency was better when the proximal anastomosis was made to the aorta and did not use the IMA as RA inflow [134]. However, Desai et al found that at one year, 21% of RA grafts going directly to the aorta had "some degree" of angiographic stenosis, which was significantly less than SVG proximal

Testing for appropriateness of RA harvesting to gauge the likelihood of hand ischemia after RA harvesting is recommended. The Modified Allen's Test (MAT) is abnormal in wide ranges (<1% to 27%), often attributed to observer variability [135]. When MAT was compared with Doppler ultrasonography of the thumb artery, MAT was noted to have a sensitivity of 100% and specificity of 97% for thumb ischemia [136]. Various adjuncts to the MAT have been proposed, including pulse oximetry, plethysmography, and Doppler ultrasonography of the hand [136-138]. However, none of these modalities have been proven to add significantly to the diagnostic accuracy of a properly performed MAT, which

Harvesting of the RA has been performed traditionally by the open, "no-touch" technique [127]. However, more recently, the trend appears to have shifted toward endoscopic RA harvesting with demonstrated functional and cosmetic advantages [139] and no increase in vasoreactivity or damaged endothelium [140]. However, harvesting the RA as a pedicle with harmonic scalpel appears to be less injurious than using electrocautery for tissue dissection

appears to accurately and safely select patients for RA harvesting.

[140]. Patency rates are similar regardless of method of harvesting [142].

artery [129], which has been corroborated by others [130].

surgery.

anastomoses [78].

Fig. 4.

Similar data have been reported by Tranbaugh et al, who compared over 800 patients undergoing CABG utilizing LIMA, RA, and SVG with matched patients undergoing CABG with LIMA and SVG only [127]. They showed significantly improved survival in the RA group in addition to improved patency of RA as compared to SVG (81% vs. 47%) in symptomatic patients who subsequently underwent diagnostic angiography after CABG (mean time to repeat catheterization 4.3 years) [127]. Importantly, RA use emerged as independent predictor of survivalat 14 years when the data were assessed by multivariable analysis [127, Figure].

Fig. 5. Comparison of Kaplan-Meier survival for propensity-matched patents (p - 0.0011, log rank test), CABG - coronary artery bypass graft surgery.

Conduct of these well-designed trials incorporating angiographic follow-up of RA grafts has provided a wealth of insight into features important to successful RA grafting. For example, as noted, early graft failures in the form of "string sign" have been noted in randomized trials [78], and several explanations for this are possible. For example, RA grafts appear

Similar data have been reported by Tranbaugh et al, who compared over 800 patients undergoing CABG utilizing LIMA, RA, and SVG with matched patients undergoing CABG with LIMA and SVG only [127]. They showed significantly improved survival in the RA group in addition to improved patency of RA as compared to SVG (81% vs. 47%) in symptomatic patients who subsequently underwent diagnostic angiography after CABG (mean time to repeat catheterization 4.3 years) [127]. Importantly, RA use emerged as independent predictor of survivalat 14 years when the data were assessed by multivariable

Fig. 5. Comparison of Kaplan-Meier survival for propensity-matched patents (p - 0.0011, log

Conduct of these well-designed trials incorporating angiographic follow-up of RA grafts has provided a wealth of insight into features important to successful RA grafting. For example, as noted, early graft failures in the form of "string sign" have been noted in randomized trials [78], and several explanations for this are possible. For example, RA grafts appear

rank test), CABG - coronary artery bypass graft surgery.

Fig. 4.

analysis [127, Figure].

notoriously susceptible to competitive flow within the native coronary circulation, as may occur when native coronary stenosis is not severe ( 70%) [128, 129]. Based on this frequent observation, use of the RA is not recommended as a conduit for CABG unless the native coronary artery stenosis is high grade ( 70%). This strategy has been associated with improved patency and reduced "string sign" in the RAPS study [78]. Radial artery graft patency is also dependent upon coronary target location, as has been noted for the RIMA graft [89]. For instance, Maniar et al found that grafts to the right coronary artery were more likely to be occluded compared with those placed to the LAD or to the circumflex coronary artery [129], which has been corroborated by others [130].

Certain inherent, unique characteristics of the RA may also contribute to "string sign" formation [131]. Limb arteries such as the RA are known to be more prone to spasm than somatic (IMA) or splanchnic arteries [132]. One key difference between the IMA and the RA is that the RA is significantly more muscular, leading to increased tendency for spasm requiring prolonged vasodilation [133]. Vasodilation of the harvested RA should begin intraoperatively by exposing the conduit to papaverine or verapamil/nitroglycerin. Verapamil/nitroglycerin may be more effective than papaverine in regards to degree of vasodilation and preservation of endothelial function, which can be a problem with papaverine (especially if injected intraluminally) and with the alpha-blocking agent phenoxybenzamine [118]. Postoperatively, most authors have recommended vasodilation with calcium channel blocking agents or long-acting nitrates for at least one month after surgery.

It is debated as to whether the proximal RA graft anastomosis should be performed to the ascending aorta or as a composite graft to other conduits. In a study of over 1,500 radial artery grafts, Maniar et al found no difference between these two types of grafting strategies [129]. Collins et al, reporting the results of the Radial Artery Versus Saphenous Vein Patency (RSVP) trial, demonstrated superior patency of RA grafts compared with SVG, and all proximal anastomoses were performed to the aorta directly [122]. Jung et al recently demonstrated using postoperative CT angiography that RA patency was better when the proximal anastomosis was made to the aorta and did not use the IMA as RA inflow [134]. However, Desai et al found that at one year, 21% of RA grafts going directly to the aorta had "some degree" of angiographic stenosis, which was significantly less than SVG proximal anastomoses [78].

Testing for appropriateness of RA harvesting to gauge the likelihood of hand ischemia after RA harvesting is recommended. The Modified Allen's Test (MAT) is abnormal in wide ranges (<1% to 27%), often attributed to observer variability [135]. When MAT was compared with Doppler ultrasonography of the thumb artery, MAT was noted to have a sensitivity of 100% and specificity of 97% for thumb ischemia [136]. Various adjuncts to the MAT have been proposed, including pulse oximetry, plethysmography, and Doppler ultrasonography of the hand [136-138]. However, none of these modalities have been proven to add significantly to the diagnostic accuracy of a properly performed MAT, which appears to accurately and safely select patients for RA harvesting.

Harvesting of the RA has been performed traditionally by the open, "no-touch" technique [127]. However, more recently, the trend appears to have shifted toward endoscopic RA harvesting with demonstrated functional and cosmetic advantages [139] and no increase in vasoreactivity or damaged endothelium [140]. However, harvesting the RA as a pedicle with harmonic scalpel appears to be less injurious than using electrocautery for tissue dissection [140]. Patency rates are similar regardless of method of harvesting [142].

Conduit Selection for Improved Outcomes in Coronary Artery Bypass Surgery 157

[13] de Canniere D, Wimmer-Greinecker G, Cichon R, et al. Feasibility, safety, and efficacy

[14] Zacharias A, Habib RH, Schwann TA, Riordan CJ, Durham SJ, Shah A. Improved

[15] Burfeind WR, Jr., Glower DD, Wechler AS, et al. Single versus multiple internal

[16] Cho KR, Kim JS, Choi JS, Kim KB. Serial angiographic follow-up of grafts one year and

[17] Nishida H, Tomizawa Y, Endo M, Kurosawa H. Survival benefit of exclusive use of in

[18] Sabik JF III, Blackstone EH, Gillinov AM, Smedira NG, Lytle BW. Occurrence and risk

[19] Tabata M, Grab JD, Khalpey Z, et al. Prevalence and variability of internal mammary

[20] Grover FL, Johnson RR, Marshall G, et al. Impact of mammary grafts on coronary bypass operative mortality and morbidity. Ann Thorac Surg 1994; 57: 559-69. [21] Galbut DL, Traad EA, Dorman MJ, et al. Seventeen-year experience with bilateral

[22] Kappetain AP, Dawkins KD, Mohr FW, et al. Current percutaneous intervention and

[23] Guru V, Fremes SE, Tu JV. How many arterial grafts are enough? A population-based study of midterm outcomes. J Thorac Cardiovasc Surg 2006; 131: 1021-8. [24] Jones JW, Schmidt SE, Miller R, Nahas C, Beall AC Jr. Suitability and durability of

[25] Gruntzig A. Transluminal dilatation of coronary-artery stenosis. Lancet 1978; 1: 263. [26] Moses JW, Stone GW, Nikolsky E, et al. Drug-eluting stents in the treatment of

[27] Tonino PAL, De Bruyne B, Pijls NHJ, et al. Fractional flow reserve versus angiography for guiding percutaneous coronary intervention. N Engl J Med 2009; 360: 213-24. [28] Pfisterer M, Brunner-La Rocca HP, Buser PT, et al. Late clinical events after clopidogrel

coronary artery bypass grafting. Circulation 2005; 112: I-299 -303.

mammary artery grafts. Ann Thorac Surg 1990; 49: 195-201.

experience. J Thorac Cardovasc Surg 2007; 134: 710-6.

practice trial. Circulation 2004; 110: II-27-35.

2004; 109: 1489-96.

114 [suppl I]: I-454-60.

2009; 120: 935-40.

486-91.

92.

Cardiol 2006; 47: 2164-71.

511-6.

of totally endoscopic coronary artery bypass grafting: multicenter European

survival with radial artery versus vein conduits in coronary bypass surgery with left internal thoracic artery to left anterior descending artery grafting. Circulation

mammary artery grafting for coronary artery bypass: 15-year follow-up of a clinical

five years after coronary artery bypass surgery. Eur J Cardiothorac Surg 2006; 29:

situ arterial conduits over combined use of arterial and vein grafts for multiple

factors for reintervention after coronary artery bypass grafing. Circulation 2006;

artery graft use in contemporary multivessel coronary artery bypass graft surgery: analysis of the Society of Thoracic Surgeons National Cardiac Database. Circulation

coronary bypass grafting procedures for three-vessel and left main coronary artery disease. Insights from the SYNTAX run-in phase. Eur J Cardio-thorac Surg 2006; 29:

multiple internal thoracic artery coronary artery bypasses. Ann Surg 1997; 225: 785-

intermediate lesions: pooled analysis from four randomized trials. J Am Coll

discontinuation may limit the benefit of drug-eluting stents: an observational study of drug-eluting verusus bare-metal stents. J Am Coll Cardiol 2006; 48: 2584-91.
