Radial Artery Grafts vs Saphenous Vein Grafts in Coronary Artery Bypass Surgery: Title and subTitle BreakA Randomized Trial FREE

Coronary artery bypass grafting (CABG) is one of the most common operations performed; the Society for Thoracic Surgery (STS) database shows that in the United States, 163 048 patients had CABG surgery in 2008. The success of CABG depends on the long-term patency of the arterial and venous grafts. The majority of patients receive left internal mammary artery grafts to the left anterior descending coronary artery and saphenous vein grafts to the remaining vessels. The improved patency of the in situ left internal mammary has led investigators to explore the use of other arterial grafts, such as the right internal or free internal mammary, the gastroepiploic, the splenic, and the radial artery, with the radial artery being the easiest to harvest. The STS database shows that 10 319 patients in the United States received radial artery grafts in 2008, suggesting that about 6% of patients undergoing CABG (10 319/163 048) have radial artery grafts.

The most important question regarding the efficacy of different conduits is their long-term patency. The Department of Veterans Affairs (VA) Cooperative Studies Program has had a long-standing interest in long-term graft patency after CABG.^{1 - 4} These studies have examined serial graft patency from 7 days up to 10 years after CABG. The trial reported here, sponsored by the VA Cooperative Studies Program, was a prospectively randomized clinical trial comparing radial artery grafts vs saphenous vein grafts in patients undergoing elective first-time CABG.

The study was approved by each site's institutional review board; written informed consent was obtained from each patient before screening. This trial studied patients undergoing elective first-time CABG that did not require concomitant valve surgery. Saphenous vein coronary bypass grafting was performed by the usual protocol for each of the study institutions. Specific details regarding study population and surgical technique can be found in the eAppendix.

Patients were assigned in a 1:1 ratio to radial artery or saphenous vein grafts using permuted blocks randomization, stratified by hospital, vessel to be bypassed (left anterior descending artery [LAD] vs the other 2), and whether surgery would be performed on pump or off pump. The surgeons decided preoperatively and prior to randomization which vessel would be the subject of the study; the criterion used was to choose the recipient coronary artery most suitable for grafting. Therapy used to improve graft patency was aspirin (325 mg) given 6 hours after surgery via the nasogastric tube. Aspirin use continued daily for at least a year. Patients were followed up every 3 months, with postoperative visits at 3, 6, and 9 months and 1 year after CABG. Patients underwent graft angiography at 1 week after surgery or before discharge from hospital and at 1 year after CABG. Patients whose study graft was occluded at 1 week or who had clinically indicated angiogram within 1 year after CABG were not asked to undergo angiogram at 1 year.

The primary end point was angiographic graft patency at 1 year after CABG, defined as any opacification of distal target by injection of the graft. The window for the 1-year angiogram was 2 to 24 months. This window was chosen to capture early clinically indicated angiograms and late selective angiograms in patients who did not have symptoms. Study grafts that were occluded at 1 week after CABG were considered occluded at 1 year. One-year graft patency data were missing if patients whose study grafts were patent at 1 week did not undergo an angiogram within the time window or if the central angiography laboratory was not able to determine graft patency.

Two hundred twelve of the 12-month angiograms for the radial artery patients and 203 for the saphenous vein patients were obtained between 10 and 14 months after surgery; 37 angiograms for the radial artery patients and 47 for the saphenous vein patients were obtained after 14 months after surgery. A sensitivity analysis was performed based on patients whose angiograms were obtained between 10 and 14 months after surgery and classifying those with occluded grafts before 10 months as occluded and those with patent grafts after 14 months as patent at 12 months. Secondary end points included myocardial infarction, stroke, repeat coronary revascularization, and death.

We estimated cardiac bypass surgery costs using the VA Decision Support System, an activity-based cost allocation system. We adjusted the costs for regional variation using the Medicare wage index, and the prior year's costs were inflated to 2009 using the general consumer price index. We analyzed costs and logged costs using linear regression with robust standard errors. Quality of life was measured at baseline, 3 months, and 12 months using the Health Utilities Index Mark III (score range, −0.37 to 1; scores less than 0 represent a state worse than death, 0 represents death, and 1 represents perfect health). Hand function was assessed with 2 standard hand-performance tests: manual dexterity was measured with a 9-hole peg test, and grip strength was measured with a dynamometer.

Angiograms were read both locally at the site and centrally in the angiographic laboratory at the Southern Arizona VA Health Care System. The readings from the central angiographic laboratory were used to determine the primary end point. All sites were required to record catheter type and size. The angiographers used the same types of catheters for the 1-week and 1-year arteriogram. The arteriograms included at least 2 orthogonal views of each graft (45° left anterior oblique and 45° right anterior oblique). Angiographers administered 100 μg of nitroglycerin into the study graft prior to the contrast injection if the patient's systolic blood pressure was greater than 90 mm Hg.

The central angiography readers were blinded as to which patients had radial artery grafts vs saphenous vein grafts. The angiograms were viewed with a Siemens-based computer system with ACOM PC and Quantcor QCA software (Siemens Healthcare, Erlangen, Germany). Using the catheter as a calibration factor, this system defined coronary diameter and percent stenosis. The reader assessed the percent stenosis at 3 sites on the graft (origin, body, and distal site). The graft was designated as occluded if the graft was 100% occluded at 1 of the sites. Qualitative features were also defined, such as lucencies, filling defects, extra luminal contrast consistent with dissection, or sharp angulations in grafts, which could contribute to occlusion.

The study was designed to have 90% power to detect 1-year patency rates of 92% in radial artery vs 83% in saphenous vein grafts, with a 2-sided type I error of 5% and an expected 1-year catheterization completion rate of 65%. Final enrollment was 733, lower than the originally planned total of 874. However, 1-year catheterization completion rate (73%) was higher than expected, and the post hoc power to detect 92% vs 83% patency was 89%.

An independent data monitoring committee reviewed the unblinded study data every 6 months. Interim analyses for efficacy were performed every 6 months starting from January 2004 with a plan to recommend study termination if the test statistic for the primary outcome crossed the 2-sided O’Brien-Fleming boundaries derived from the α spending function.⁵

All analyses were performed according to intention to treat. The primary end point, study graft patency at 1 year after CABG, was compared between the 2 groups using a χ² test. Multiple logistic regression analysis was used to adjust for stratification factors and for characteristics that were potentially predictive of graft patency, including size of the recipient vessel (≤2 mm vs >2 mm), diabetes mellitus (presence/absence), smoking history (current smoker vs other), level of low-density lipoprotein cholesterol, and quality of study graft (good/poor). We also performed prespecified subgroup analyses and assessed treatment × subgroup interaction. Sex was not included in the subgroup analysis or the logistic regression because the number of women was small.

For baseline characteristics and secondary end points, χ² tests were used to compare binary or categorical variables; 2-sample t tests or Wilcoxon rank sum tests were used to compare continuous variables. To account for missing primary end point data, we assessed potential angiographic bias by comparing baseline characteristics between those patients who underwent the 1-year catheterization with those who did not, both overall and within each treatment group. We also performed as-treated and per-protocol analyses and multiple imputations as sensitivity analyses to the intent-to-treat analysis on primary end point. Analyses were performed using SAS version 8.2 (SAS Institute, Cary, North Carolina). A 2-sided P < .05 indicated statistical significance.

From February 2003 through February 2008, we randomized 757 patients from 11 VA medical centers (eTable 1). With 24 postrandomization exclusions, 733 patients (366 in the radial artery group, 367 in the saphenous vein group) were included in the analysis (Figure 1). Demographic and clinical characteristics of the patients were similar in the 2 groups (Table 1).

There was a difference in the proportion of the 2 treatment groups receiving what they were randomized to receive: 332 of 366 patients randomized to receive a radial artery graft received one (91%; 95% confidence interval [CI], 87%-93%), while 362 of 367 patients randomized to receive a saphenous vein graft received it (99%; 95% CI, 97%-100%) (Table 2). (Refer to eAppendix for reasons.) The surgery was performed on pump for 644 of 733 patients (88%: radial artery, 325/366; 89%; 95% CI; 86%-92%; saphenous vein, 319/367; 87%; 95% CI, 83%-90%). Eighteen patients in the radial artery group (5%; 95% CI, 3%-7%) and 72 patients in the saphenous vein group (20%; 95% CI, 16%-24%) underwent endoscopic vessel harvesting (P < .001).

The 30-day surgical mortality was 0.7% (95% CI, 0.2%-1.6%; 5/733). The 5 deaths within 30 days of surgery included 1 cardiac death, 1 sudden death, 1 due to infection, 1 due to postoperation complications, and 1 for an unknown reason. Surgical complications were similar; the only difference was that patients with radial artery grafts had fewer hemorrhages (defined as needing blood transfusions) than patients with saphenous vein grafts. There were fewer total surgical complications in patients who underwent endoscopic vessel harvesting than those who underwent other vessel harvesting (endoscopic, 2/90; 2%; 95% CI, 0%-5%; other, 82/643; 13%; 95% CI, 10%-15%; P = .001).

Seventy-six percent of patients (560/733) had the 1-week angiogram at a median time of 7 days after surgery (range, 1-141 days), with 328 of the angiograms (58%) studied within 7 days of surgery. Time until 1-week angiogram was not significantly different between the 2 groups (P = .54). There was no difference in 1-week patency between patients who received radial artery grafts (285/288; 99%; 95% CI, 97%-100%) vs saphenous vein grafts (260/267; 97%; 95% CI, 95%-99%), nor between patients who underwent endoscopic vessel harvesting (61/63; 97%; 95% CI, 96%-100%) vs other vessel harvesting (484/492; 98%; 95% CI, 97%-99%).

Five hundred thirty-three patients (73%) had 1-year angiograms. The median time until the 1-year angiogram was 371 days (range, 61-725 days), with 415 (78%) of the angiograms studied between 10 and 14 months of surgery (34 before 10 months, 84 after 14 months). One-year angiograms performed between 2 and 9 months after surgery were more likely to be clinically indicated and had a lower patency rate (22/28; 79%; 95% CI, 63%-94%) than those performed later in the time window (10-14 months, 377/415; 91%; 95% CI, 88%-94%; after 14 months, 78/82; 95%; 95% CI, 90%-100%). Time until 1-year angiogram was not different between the 2 groups (P = .31). The central laboratory determined graft patency for 525 of 533 1-year angiograms; 8 angiograms could not be interpreted because the study vessel was not imaged. Assigning the 10 study grafts that were occluded at 1 week after CABG as occluded at 1 year, the primary outcome was available for 73% of patients (535/733).

There was no significant difference in 1-year graft patency between radial artery (238/266; 89%; 95% CI, 86%-93%) and saphenous vein grafts (239/269; 89%; 95% CI, 85%-93%; unadjusted odds ratio [OR], 1.07; 95% CI, 0.62-1.84; P = .82) (Table 3). The results remained similar in as-treated analysis (adjusted OR, 0.87; 95% CI, 0.50-1.53; P = .63), per-protocol analysis (adjusted OR, 0.92; 95% CI, 0.52-1.62; P = .78), by multiple imputations (adjusted OR, 0.92; 95% CI, 0.56-1.51; P = .73), and in the sensitivity analysis on time window (adjusted OR, 1.08; 95% CI, 0.60-1.93; P = .80). Graft patency was similar between the 2 groups in grafts to different coronary arteries (LAD/circumflex/right coronary artery) (Table 3). In retrospect, given that 100% occlusion occurred less frequently than expected, the primary end point analysis may be underpowered to detect a difference. After adjusting for stratification factors, including site, target vessel (LAD vs other), and whether the surgery was performed on pump vs off pump, the adjusted OR was 1.04 (95% CI, 0.60-1.80; P = .90). After further adjusting for smoking (current smoker vs other), which was associated with lower 1-year patency rate in univariate analysis, the adjusted OR was 0.99 (95% CI, 0.56-1.74; P = .98). For the prespecified subgroup analyses shown in Figure 2, a significant interaction with graft type was indicated for diabetes (P = .04); radial artery grafts had lower patency rate than saphenous vein grafts in patients with diabetes and the reverse was true in patients without diabetes.

The primary study outcome is a binary one, graft occlusion or not at 12 months. As described in the “Methods” section, the angiography reader assessed the percent stenosis at 3 sites on the graft (origin, body, and distal site). If one reader noted the maximum of the percent stenosis at the 3 sites, then another reader could examine the development of disease in the study grafts at 1 year (Table 3). There was a higher incidence of a 99% occlusion (a “string sign”) with radial artery grafts (21/266; 8%; 95% CI, 5%-11%) compared with saphenous vein grafts (3/269; 1%; 95% CI, 0%-2%; P < .001). The incidence of severe stenosis (75%-100%) was significantly higher in radial artery grafts (63/266; 24%; 95% CI, 19%-29%) than in saphenous vein grafts (43/269; 16%; 95% CI, 12%-20%; P = .03), but there was no evidence of progression of disease in target vessels. Although internal mammary artery (IMA) patency was not a study end point, we do have data on 458 left IMA grafts at 1 year with a patency rate of 96% (438/458). The patency rate was similar in the 2 groups (saphenous vein, 97% [221/228]; radial artery, 94% [217/230]; P = .18).

There was no significant difference between the 2 groups in the secondary outcomes, including death, myocardial infarction, stroke, and repeat coronary revascularization (Table 3). There was no difference in the number and types of adverse events, including serious adverse events, which occurred in 330 of 733 patients (45%: radial artery, 158/366; 43%; 95% CI, 38%-48%; saphenous vein, 172/367; 47%; 95% CI, 42%-52%; P = .31). Sixteen patients died within 1 year of CABG (9 in the radial artery group, 7 in the saphenous vein group; P = .61), including 3 cardiac deaths (1 in the radial artery group, 2 in the saphenous vein group).

Use of statins, aspirin, and β-blockers was similar between the 2 groups. Use of calcium-channel blockers was higher in the radial artery group (saphenous vein, 70/367; 19%; radial artery, 162/366; 44%; P < .001) (eTable 2). One-year graft patency was not different between patients who used or did not use calcium-channel blockers in the saphenous vein group (used calcium-channel blockers, 50/57; 88%; 95% CI, 79%-95%; did not use calcium-channel blockers, 189/212; 89%; 95% CI, 85%-93%; P = .76) or in the radial artery group (used calcium-channel blockers, 108/116; 93%; 95% CI, 88%-97%; did not use calcium-channel blockers, 130/150; 87%; 95% CI, 81%-92%; P = .09). Ninety-three percent of patients (684/733) were given perioperative vasodilators or inotropes (eTable 3).

In the saphenous vein group, the 1-year patency was lower in patients who underwent endoscopic vessel harvesting than in those who underwent other vessel harvesting (endoscopic, 40/51; 78%; 95% CI, 67%-90%; other, 199/218; 91%; 95% CI, 88%-95%; P = .009). The difference was not significant in the radial artery group (endoscopic, 11/11; 100%; 95% CI, 72%-100%; other, 227/255; 89%; 95% CI, 85%-93%; P = .61).

In the saphenous vein group, the 1-year patency rate was higher in patients whose surgery was done on pump (on pump, 214/237; 90%; 95% CI, 87%-94%; off pump, 25/32; 78%; 95% CI, 64%-92%; P = .04). The difference was not significant in the radial artery group (on pump, 221/247; 89%; 95% CI, 86%-93%; off pump, 17/19; 89%; 95% CI, 76%-100%; P = .99).

There was no difference in cost; the mean (SD) cost of the hospital stay for a patient with saphenous vein graft was $47 560.30 ($38 523.10), and the mean (SD) cost for a radial artery patient was $49 390.40 ($30 352.70) (P = .51). There were no significant differences in Health Utilities Index scores at 3 months (mean [SD] scores, saphenous vein, 0.67 [0.29]; radial artery, 0.64 [0.27]; P = .40) or 12 months (saphenous vein, mean [SD], 0.67 [0.29]; radial artery, 0.64 [0.27]; P = .23). Hand function measures at 12 months showed minimal decrease in grip strength in the radial artery group (mean difference between the arm where radial artery graft was harvested and other arm: before surgery, −4.9; 95% CI, −6.1 to −3.6; 12 months after surgery, −6.0; 95% CI, −8.6 to −3.5) and no difference in 9-hole peg test results (mean difference between the arm where radial artery graft was harvested and other arm: before surgery, 1.0, 95% CI, 0.5-1.4; 12 months after surgery, 0.9; 95% CI, 0.0-1.9).

One year after CABG, there was no difference in angiographic patency between radial artery grafts and saphenous vein grafts in men. Although most clinicians assume that compared with vein grafts, arterial grafts have an improved patency rate, there are little multi-institutional prospective data on radial artery graft vs saphenous vein graft patency.^{6 - 11} To date, the best data come from the 2004 report of the RAPS Study, which was funded by the Canadian Institutes of Health Research. This prospective randomized multicenter trial reported improved 1-year patency of radial artery grafts compared with saphenous vein grafts (92% vs 86%).⁶ These results are different from ours because we saw no difference in patency at 1 year. One possible reason for this difference is the improved saphenous vein graft patency in our study compared with the RAPS study (89% vs 86%). Another possible reason is that in the RAPS study, each patient had both study grafts (a radial artery and at least 1 saphenous vein) while in our study, each patient had only 1 study graft with the surgeon choosing the best recipient vessel for the study graft. In theory our approach should have resulted in the best chance for patency for each graft and an opportunity to define the patient characteristics that predict graft patency. Because each patient had only 1 study graft, we are also able to separate the complications associated with radial artery vs saphenous vein harvesting. With catheterizations at 1 week and 1 year after surgery, we can also separate the perioperative technical surgical problems from the pathophysiology of disease progression over 1 year and have a known denominator to define 1-year patency. Our data show no difference in surgical or postoperative complications between the grafts.

It is important to acknowledge that despite similar patency rates, there was more disease in the radial artery grafts at 1 week and at 1 year. Although the explanation for this difference is not clear, there is clearly more spasm or string sign in radial artery grafts. One of the most important factors determining radial artery graft patency and development of string sign is the degree of stenosis in the native target vessel. In our study we used more than 70% proximal stenosis as the entry criterion for the study vessel. It is possible that vessels with more severe proximal stenosis would have better radial artery graft patency rates. It is important to note that this study was conducted predominantly in men; to our knowledge, there is no literature that focuses on radial artery grafts in women.

Our data show no differences in radial artery graft patency but show lower vein graft patency when endoscopic harvesting was used. Although our number of patients is too small to provide a definitive answer, these results are consistent with a report from the PREVENT IV study¹² that showed endoscopic vein harvesting was detrimental in vein graft patency. Our data show no significant difference with on-pump vs off-pump surgery for radial artery graft patency but show higher vein graft patency on pump, consistent with a recent report from the ROOBY trial.¹³

The radial artery can develop atherosclerosis and calcification, especially in patients with diabetes. Although radial artery grafts in diabetic patients were found more likely to be patent at 1 year after coronary artery bypass compared with saphenous vein grafts,⁹ our study found the opposite to be true. Other studies of complete arterial revascularization appear to favor arterial grafts in diabetic patients despite this propensity for atherosclerosis,¹⁴ while in a recent trial, the use of a radial as a second arterial conduit as opposed to vein grafting did not confer a survival benefit in diabetic patients.¹⁵ This unexpected result is perhaps related to the augmented radial vasoreactivity and atherosclerotic disease characteristic of patients with diabetes. These findings indicate that the radial artery conduit advantage demonstrated by others in the general CABG population may primarily be in patients who do not have diabetes. Longer follow-up will be necessary to determine whether radial artery grafts are truly of benefit in diabetic patients.

Among patients undergoing first-time elective CABG, the use of a radial artery graft compared with saphenous vein graft did not result in greater 1-year patency. Because the important question is long-term patency, the VA Cooperative Studies Program has funded a 5-year angiographic follow-up of these patients to define chronic graft patency in this population.