The success of coronary artery bypass grafting (CABG) is dependent on the long-term patency of the arterial and venous grafts. Worldwide, more than 800,000 patients undergo CABG annually, with more than 350,000 patients operated each year in the U.S. The majority of these patients receive left internal mammary artery (IMA) grafts to the left anterior descending (LAD) coronary artery and saphenous vein grafts (SVGs) or other conduits to the remaining vessels. Based on small studies of selected groups of patients, it is generally believed that SVGs have a 40% to 50% 10-year patency and that the IMA has a 90% to 95% 10-year patency (1- 7). To address the question of long-term patency of saphenous vein and IMA grafts, the Department of Veterans Affairs Cooperative Studies Program supported a 10-year angiographic follow-up study of patients who were participants in two VA Cooperative Trials in the 1980s. These trials were originally designed to examine the effect of antiplatelet agents on graft patency after CABG with postoperative coronary angiography one week, one year, and in some cases three years after surgery (8- 12).

This trial, organized and funded by the Cooperative Studies Program of the Department of Veterans Affairs Research and Development Service, obtained data from 1,254 male patients entered into two studies at 13 participating hospitals from July 1983 to September 1988. The Institutional Review Board approved each study, the subjects gave written informed consent, and procedures were followed in accordance with institutional guidelines (8- 13).

Coronary artery bypass grafting was done in a standard fashion as previously described (8- 12). The decision to use arterial or venous conduits was made by the attending surgeon. The patient was eligible for the trial only if the determination was made preoperatively to utilize at least one SVG.

The angiographic analysis was identical to our earlier trials (8- 12). Briefly, the left IMA and each aortic anastomosis were selectively engaged and injected. When the status of a vein graft could not be determined by graft or stump injection, an aortic root angiogram was performed. Selective angiography of the native coronary arteries was performed during the one-week and one-year catheterization only when a graft was occluded. Selective angiography of the native coronary arteries was routinely performed at the 3-year and 10-year catheterization. Graft patency and stenosis were defined in a central angiographic laboratory with a computer-based system (8- 12). The data are presented as time-based occlusion rates where occlusion is defined in two ways. The primary event of interest is 100% stenosis (Figures 1, 2, 3, 4). The secondary event is 50% to 99% stenosis, not including vessels totally occluded (Figure 5). Because of long-term follow-up and the study group being composed of middle-aged men with coronary artery disease, some patients died before the completion of the study. We obtained autopsies in as many patients as possible and included the autopsy patency data in our analysis.

Traditionally, reports of long-term graft patency rates have used coronary angiography results from a single time point after CABG. Patency rates from a single time period distant from the original operation may result in biased estimates if there is no accounting for interceding interventions.

The data from this study posed two major analytic problems. The first problem was that the exact time of graft occlusion could not be known. We addressed this by using interval-censored observations in the survival analysis (PROC LIFETEST in SAS [SAS Institute Inc. SAS/STAT Version 8, Cary, North Carolina]). This analysis requires identification of the time interval in which the occlusion occurred. If a patient had only one angiogram and this showed an occluded graft, the time between the date of the CABG and date of angiography was used as the occlusion occurrence interval. If a patient had multiple angiograms, the time between the date of the most recent angiogram that showed a graft was patent and the date of the angiogram that showed occlusion was used as the occlusion occurrence interval. The date of the latest angiogram was used as the right censored time for grafts, which remained patent. For patients who had patency data from an autopsy, the date of death was used. If there was an intervention on a stenotic but still patent graft by either percutaneous coronary intervention or repeat surgical revascularization, the graft was right censored on the date of the angiogram and no further data were collected for that graft. Comparisons of Kaplan-Meier product-limit survival curves were made with the log-rank test.

Although time-related analyses of graft patency data used the exact date of each postoperative angiogram, for convenience of presentation, some information is presented in arbitrarily defined time frames. The 1-week data included catheterizations performed between 1 and 60 days post-operation, the 1-year included catheterizations performed between 61 days and 1.5 years post-operation, the 3-year included catheterizations after 1.5 to 4.5 years post-operation, the 6-year included catheterizations after 4.5 up to 8 years post-operation, and the 10-year included catheterizations after 8 to 12 years post-operation.

The second analytic problem was that there were multiple grafts (i.e., clustered observations, within a patient). We showed previously (14) that graft patency within a patient is not independent. This does not affect the estimates for patency rates, but it does cause the standard error terms to be underestimated. Recently, the SAS macro IWM (16) has become available for the analysis of clustered, interval-censored survival data (15- 16). This approach produces robust estimates of the standard error terms by adjusting for the correlated nature of the clustered observations. Patient-related risk variables, graft-related risk variables, and CABG processes of care variables were used as candidate independent variables in the IWM macro to identify the set of variables that jointly predict 10-year graft patency. Variables that were significant at the p < 0.10 level were included in the final models. The Appendix gives a description on how to interpret the importance of the variables.

The t test and chi-square test were used to compare pre-CABG patient risk factors for patients who received a 10-year catheterization versus those who did not.

Of the 1,254 patients undergoing CABG operations during the study period, 1,079 patients (86%) had at least one postoperative cardiac catheterization. This included 1,074 patients with one or more SVG and 457 patients with an IMA graft. Two hundred twenty-five patients had a single postoperative catheterization (18%), 328 patients had two catheterizations (26%), 384 had three catheterizations (31%), 119 had four catheterizations (9%), and 23 had five catheterizations (2%). Patient status during each of the five angiographic periods is given in (Table 1). A total of 919, 799, 367, 170, and 369 patients had cardiac catheterizations nominally at the 1-week, 1-year, 3-year, 6-year, and 10-year periods, respectively. The median times to the catheterizations were 8 days and 1.0, 3.0, 6.2, and 9.7 years, respectively. Ninety-six and sixty-seven percent of the 6- and 10-year catheterizations, respectively, were performed for clinical, not study-related, reasons. The 10-year catheterizations occurred between May 1991 and January 1999. There were 523 deaths before the 10-year catheterization. Of these 523 patients, serial catheterizations or autopsy data were obtained on 396 patients (76%). One hundred thirty patients had a single catheterization before death (25%), 168 had two catheterizations (32%), 87 had three catheterizations (17%), and 11 had four catheterizations (2%). There were autopsy data with graft patency defined on 71 patients. The details of the patient data, including clinical outcomes and complications of the previous catheterizations, have been reported (8- 12). The clinical characteristics of patients with SVGs and IMA grafts (Table 2) are presented. The patients who did not undergo the 10-year catheterization had a higher incidence of diabetes (p = 0.022) and a trend toward more hypertension (p = 0.057) and a higher incidence of left main disease (p = 0.084) (Table 3).

The 10-year patency for SVGs was 61% (Figure 1). If a vein graft was patent at the 1-week study, the 6-year patency rate was 76% and the 10-year patency rate was 68% (Figure 2). We note that the sudden drop in percentage patency in the graphs around six months post-CABG occurs because a large amount of new information was obtained at the one-year angiogram and hence should not be viewed as having physiologic relevance. For IMA grafts, the 10-year patency was 85% (Figure 1). If an IMA graft was patent at 1 week, the 6-year and the 10-year patency rates were 90% and 88%, respectively (Figure 2).

The data for a specific vessel grafted at 10 years show that vein graft patency to the LAD (69%) was better (p < 0.001) than vein graft patency to the right coronary artery (RCA) (56%) or circumflex (CX) (58%) (Figure 3). There was no difference in the recipient vessel diameters among SVGs to the LAD (1.67 ± 0.43 mm) versus the CX (1.68 ± 0.44 mm) versus the RCA (1.71 ± 0.50 mm), p = 0.179. If the vein graft to the LAD was patent at 1 week, the 10-year patency was 74%. Similar improvement was seen at 10 years with early patency for the RCA (64%) and for the CX (66%). Left IMA grafts to the LAD had a higher patency than single SVGs to the LAD (85% vs. 69%, p < 0.001).

Along with the location of the distal site, the best predictor of graft patency over the 10-year post-bypass period was the diameter of the recipient vessel by angiographic measurement (Figure 4). In vessels >2.0 mm the 10-year patency was 88%, versus 55% in vessels with diameters ≤2.0 mm (p < 0.001). For SVGs to the LAD, the 10-year patency was 90% for vessels >2.0 mm versus 52% for vessels ≤2.0 mm (p < 0.001). For the IMA, the 10-year patency was 100% for vessels >2.0 mm versus 82% for vessels ≤2.0 mm (p = 0.008). After controlling for hospital effects, other positive significant predictors of graft patency over the 10-year post-bypass period were older age, use of aspirin after CABG surgery, lower serum cholesterol, and lowest Canadian Functional Class (Table 4). If a graft was patent at one week, significant positive predictors of graft patency rates over the 10-year post-bypass period included IMA grafts to the LAD, larger diameters of the recipient vessel, older age, and lower serum cholesterol (Table 5). Lower platelet count was mildly predictive (p = 0.089) of graft patency. Interestingly, the presence of diabetes requiring insulin and cigarette smoking did not predict graft patency independent of whether the vein graft was patent early or not. Other variables that were tested in the model that were not predictive of vein graft patency included the following: race, number of diseased vessels, presence of peripheral edema, systemic emboli, hypertension, white blood cell count, prior myocardial infarction, proximal anastomotic site, temperature of cardioplegia solution, whether preservation solution given via graft or not, heart size, use of positive inotropic, vasodilator, antiarrhythmic agent perioperatively, use of an intraaortic balloon, lowest body temperature, and total volume of cardioplegia.

The primary end point of this study was the graft patency at 10 years, but disease clearly develops in grafts. In SVGs, which developed 50% to 99% stenosis by 10 years, there was no difference among grafts to the LAD, CX, and RCA (17% to 18%). There is a significant difference in IMA grafts with 50% to 99% stenosis compared with SVGs to the LAD (5% vs. 22%, p < 0.001) and with IMA grafts compared with all SVGs (5% vs. 18%, p < 0.001). The difference appears to start somewhere after three years postoperatively (Figure 5).

These data report 10-year serial angiographic follow-up of patients operated on in the 1980s. For SVGs, the 10-year patency is 61%, and if a SVG is patent at 1 week after surgery, that graft has a 68% chance of being patent at 10 years. For IMA grafts, the 10-year patency is 85% and if an IMA graft is patent at 1 week, that graft has a 88% chance of being patent at 10 years. Combining the data in this report with our previous work (8- 12), we know that serial SVG patency is 95% at 1 week, 84% at 1 year, 80% at 3 years, 69% at 6 years, and 61% at 10 years (Figure 1). The serial data for IMA grafts are 99%, 95%, 93%, 87%, and 85% at each of the same time points. Although the long-term patency of IMA grafts is better than that for SVGs, the 10-year patency of SVGs is better than what has been previously reported, and the 10-year patency of IMA grafts is somewhat worse than previously thought.

The most often quoted results in published reports regarding graft patency after CABG are that at 10 years, SVGs have about a 40% to 50% patency and that IMA grafts have a 90% to 95% 10-year patency. The initial studies that are the basis of this information were not prospective and reported on selected patients operated upon in the 1970s (1- 6). In these retrospective studies, graft patency was not determined on everyone in the initial cohort, but rather on those patients who received angiography most often because of symptoms. In 1996 there was a report from a larger database, which included a combination of first surgery and re-operation surgery (7). Even though that trial was designed to follow patients prospectively, the authors state that their plans “were thwarted by lack of funds” and their graft patency data, in the subset that underwent catheterization, were similar to the numbers cited above. When the present study was planned, the longest angiographic follow-up studies of IMA grafts were from the Cleveland Clinic (6,17- 18), Karolinska Hospital (19), and the Montreal Heart Institute (3,20). At or past 10 years, each of these centers has reported studying 37, 39, and 20 patients, respectively. Although these reports are important and are obviously responsible for the enthusiasm for using IMA grafts today, they are relatively small retrospective studies. Perhaps the most telling criticism of these studies is that the type of patients operated on 10 to 15 years ago is not the same as patients undergoing CABG now. In the report from the Karolinska Hospital, the 37 patients studied at 11 years had an 88% IMA patency rate (19). The baseline characteristics of this patient population included an average ejection fraction of 64% and 42 of 99 (42%) of the original patients had only IMA grafts. This obviously does not represent the current cardiac surgical population. The patients in our study had average ejection fractions of 61%, but they received 2.9 vein grafts per patient. Another reason why the earlier data may not be applicable now is that the patient population operated on today has more extensive disease than those operated on in the 1970s because today surgeons must operate on patients who have been denied percutaneous coronary intervention or are having second and third revascularization procedures.

All of the studies from the 1970s were done prior to the widespread use of antiplatelet therapy after CABG, and it is now well established that aspirin helps to prevent SVG occlusion (8- 12). In the 1970s, very little attention was paid to risk factor control and aggressive lowering of low-density lipoprotein cholesterol, which may alter SVG patency and improve clinical outcomes after CABG (21- 23). Lastly, surgical techniques have been modified with regard to preparation of the vein graft before implantation; for instance, we know that cold vein graft preservation solution improves long-term vein graft patency (13).

Comparisons are often made between veins and arteries used as conduits, with the general belief that arterial conduits have a better long-term patency. In fact, the IMA has been proclaimed the graft of choice to bypass the LAD because of its presumed excellent long-term patency rate. A 1986 editorial in the New England Journal of Medicine (23) stated, “at 10 years internal mammary artery grafts remained in excellent condition; nearly 95% were patent, with no signs of deterioration.” Although this suggests that the IMA is an ideal conduit, it is important to emphasize that there are no randomized prospectively controlled trials comparing IMA to SVGs. In the past surgeons tended to use IMA grafts in better candidates (i.e., the patient was clinically stable, without severe lung disease, diabetes, and so on). This clinical approach is borne out by our data and that of others, which show that patients receiving SVGs tend to have more severe disease initially than the patients receiving IMA grafts (9- 10,13). This argument is probably not germane today, when the characteristics of patients have dramatically increased toward more complex disease as a result of percutaneous coronary intervention, and yet IMA grafts are still used in the majority of patients. In the present study, where patients were not randomized to different conduits, the 10-year patency of left IMA grafts to the LAD is 85%, as opposed to 69% for SVGs to the LAD. Another argument that is often used to support the IMA as the conduit of choice is that retrospective studies have shown that survival is improved in patients receiving IMA grafts as opposed to vein grafts (4,6,25- 26). Even though these studies adjust for demographic and clinical differences, they are still retrospective nonrandomized analyses. Data from our study group have shown that use of the IMA is associated with a longer operative time as well as increased postoperative bleeding (27). Thus, the sicker, more complicated patients 10 to 15 years ago would have received vein grafts and not IMA grafts.

The data on predictors of graft patency are interesting. We and other investigators have previously shown that the size of the recipient vessel is an important predictor of graft patency (13). The simple explanation has been that the larger the recipient vessel, the better the flow and distal runoff. The size of the recipient vessel is also thought to be the explanation for why SVGs to the LAD do better than those to the right or CX ((Table 4) and (Table 5), Figure 3). The other significant patient specific predictors of graft patency are, for the most part, a reflection of the disease in the patient (i.e., there was worse SVG patency in younger patients, which probably is a reflection of worse native coronary disease in younger patients, and patients with elevated serum cholesterols). Interestingly, diabetes and cigarette smoking were not predictors of poor graft patency. We have previously reported on the beneficial effects of aspirin on SVG patency up to one year after surgery (11- 13). That observation is consistent with our present data that show the use of aspirin is predictive when looking at all grafts (Table 4), but not in grafts that were patent originally at one week (Table 5). Our earlier work also showed that the temperature used to preserve the vein graft was important, with improved three-year SVG patency when colder temperatures were used (13). We hypothesized that the colder temperatures and the type of preservation fluid used in the vein might have preserved endothelial function and thus improved long-term graft patency.

We believe that the estimates of 10-year graft patency presented here are less biased than any results previously published. First, although it is generally believed that patients who die may have much higher occlusion rates than survivors, the patency rate from 71 autopsies in our study (13.6% of all deaths) was 76% (156 of 205 grafts). Therefore, the effect of lost observations due to death on long-term patency rates may be minimal. Second, although only 33% of the 10-year catheterizations were done strictly for the purposes of this study, this is a substantial proportion compared with previous reported 10-year data, which used retrospective data collection and based their results solely on symptomatic patients. The use of survival analysis techniques to estimate long-term patency is a third technique in fine-tuning the estimates of graft patency. Angiography at any time point can only tell that a graft has become occluded at that point in time, but it cannot determine when the occlusion occurred. Thus, to use the time of the angiography to estimate graft patency will overestimate the patency rate. The use of interval censoring is an attempt to more accurately estimate the time that the occlusion occurred. Lastly, as in all our previous reports, we acknowledge that graft patency within an individual is correlated and we have adjusted for this association.

It may be reasonable to assume that graft patency influenced survival. The data are only from the survivors who, by definition, may have had better graft patency than the nonsurvivors. We acknowledge that not being able to address whether graft patency influenced survival is a potential weakness in our study, but this would be true for any long-term study of graft patency after CABG. Owing to the long-term nature of the study and the fact that coronary artery disease is a progressive disease, we used data from clinically indicated angiograms, in part, for the 6- and 10-year follow-up. In this older patient population it is difficult to get patients to agree to repeat invasive procedures when they already have had a clinically indicated catheterization, are asymptomatic, or when the attending cardiologist would not consider any interventional procedure. Two notes of caution: the patients who did not undergo the 10-year catheterization had more diabetes and hypertension and, therefore, the current estimates of patency may be overestimated. Lastly, all patients were men and, therefore, long-term angiographic follow-up of CABG grafts is not available for a female cohort.

In conclusion, we report long-term analysis of SVG and IMA patency for 10 years after operation, showing that patency for IMA grafts is better than for SVGs, but the 10-year patency for SVGs is better and 10-year patency for left IMA grafts is worse than previously thought. The most important predictors of long-term graft patency are initial patency at one week after CABG and the diameter of the recipient vessel into which the graft is placed.

For the Statistical Appendix and a list of the Department of Veterans Affairs Study Group members, please see the December 7, 2004, issue of JACC at http://www.onlinejacc.org.

Long-Term Patency of Saphenous Vein and Left Internal Mammary Artery Grafts After Coronary Artery Bypass Surgery: Results From a Department of Veterans Affairs Cooperative Study