The surgeon’s use of saphenous veins and a variety of arterial conduits to bypass obstructive coronary atherosclerotic disease preceded percutaneous revascularization by about a decade. In that relatively short period of time, the inferiority of venous compared to arterial grafts and the potential for atheroembolization with vein graft manipulation at surgery had become apparent. When Andreas Gruentzig reported the first 50 patients treated with percutaneous coronary angioplasty in 1979, 5 had undergone saphenous vein graft (SVG) dilatation, and 3 (60%) had developed restenosis, leading him to surmise “the different kind of disease may explain the high incidence of recurrence in graft stenosis” and to question the wisdom of percutaneous SVG intervention.¹ In the more than 35 years since that observation, interventional cardiologists have struggled with the indications for SVG intervention because of higher acute complications, more restenosis than was observed in native coronary arteries and arterial grafts, rapid disease progression in nontarget sites, and high late cardiac event rates. By 1983, outcomes of SVG intervention had been more completely characterized, and the first left internal mammary artery intervention had been reported.² In a subsequent summary of several thousand reported cases of balloon angioplasty in SVGs, procedural mortality was less than 1%, Q-wave infarction occurred in less than 2% of cases, and emergency surgery was required in 0.3% to 4% of cases.³ A gradient was observed in SVG restenosis rates, with very high rates approaching 70% at proximal anastomoses and progressively lower rates in more distal locations (see discussion mid-SVG and distal anastomosis). Subsequent maturation of graft percutaneous coronary intervention (PCI) occurred with improved understanding of patient and lesion selection, application of stents and embolic protection strategies, prediction and prevention of complications, and use of intravascular imaging as discussed below. Unfortunately, the goal of procedural and long-term safety and optimal durability of graft PCI has been illusive following treatment with SVGs.

Use of stents in SVG intervention was supported by the report in 1995 of the Palmaz-Schatz multicenter registry experience in over 500 patients with the following favorable outcomes: procedural success, 97%; stent thrombosis, 1.4%; in-hospital mortality, 1.7%; urgent surgery, 0.9%; and restenosis in 18% of de novo lesions and 46% of restenotic lesions.⁴ In the Saphenous Vein De Novo (SAVED) trial, 220 patients with SVG stenosis of at least 50% and angina or objective evidence of ischemia were randomized to deployment of a Palmaz-Schatz stent or balloon angioplasty.⁵ Patients with lesion lengths requiring more than 2 stents, myocardial infarction within 7 days, or graft thrombus were excluded. Patients who underwent stent implantation had a higher procedural efficacy, defined as a reduction in stenosis to less than 50% of the vessel diameter with the assigned therapy (92% vs 69%; P < .001). Patients assigned to stents had a larger increase in lumen diameter immediately (1.92 vs 1.21 mm; P < .001) and a greater net gain in lumen diameter at 6 months (0.85 vs 0.54 mm; P = .002). There was more bleeding in stented patients due to warfarin anticoagulation. Major in-hospital complications were otherwise similar in the 2 groups, although there was a trend toward fewer non–Q-wave myocardial infarctions in the stent group. This observation led some operators to investigate a strategy of direct stenting in SVGs, attempting to minimize distal atheroembolization.⁶

In the SAVED trial, restenosis occurred in 37% of stented patients and in 46% of patients treated with balloon angioplasty (P = .24). Event-free survival at 240 days (freedom from death, myocardial infarction [MI], or repeat revascularization) was significantly higher in stented patients (73% vs 58%; P < .03). It was noted that late lumen loss was significantly greater in patients who had high-pressure stent deployment (≥16 atm), suggesting that high-pressure stent expansion may be undesirable in SVGs. Less aggressive stent expansion was also supported by an intravascular ultrasound study of over 200 SVGs that showed that stent expansion to ≥100% of the reference cross-sectional area was associated with significantly more non–Q-wave infarctions (29% vs 17%; P = .05) and similar rates of target vessel revascularization at 1 year (31% vs 26%; P = .3).⁷

Observational reports of intervention in SVGs suggested that outcomes were somewhat better in 1995 to 1998 compared to earlier years (1-year event-free survival of 71% vs 59%; P < .001) and that use of stents had a protective effect.⁸ However, in an analysis of over 600 SVG patients in 5 large randomized trials, major adverse cardiac event (MACE) rates were twice that observed in native-vessel intervention.⁹

In the past decade, the importance of atheroembolic MI complicating PCI has become increasingly apparent, especially in SVG interventions. In relatively simple SVG lesions, stenting resulted in MI in 15% to 20% of patients. The rate of MI increased with lesion complexity, length, and estimated plaque volume,¹⁰ and in approximately 1000 patients who underwent SVG intervention, creatine kinase (CK)-MB elevation was the best independent predictor of late mortality.¹¹ Strategies to prevent atheroembolic infarction have evolved over the years from occlusion aspiration to the use of filters. The first embolic protection device (EPD) method tested was the PercuSurge Guardwire system (Medtronic, Dublin, Ireland), which used a 0.014-inch guide wire with a compliant distal occlusion balloon. Inflation of the guide wire balloon interrupted blood flow, and stent deployment was followed by aspiration and balloon deflation. Use of this system in 24 SVG stent procedures was found to prevent infarction in 23 patients, and 95% of aspirates had typical atherosclerotic debris.¹² However, it was the Saphenous Vein Graft Angioplasty Free of Emboli Randomized (SAFER) trial that unequivocally and singularly demonstrated in a randomized study the value of embolic protection during SVG intervention. In 801 patients randomized to PercuSurge Guardwire or unprotected stent implantation, 30-day MACE was reduced by 42% with use of the Guardwire (from 16.5% to 9.6%; P < .001, primarily due to reduced infarction from 14.7% to 8.6%; P = .008).¹³ In SAFER, use of EPD was demonstrated to be beneficial over a wide range of lesion lengths (even very short lesions), and this strategy was shown to be cost effective. Advantages of this technique included its low profile, which permitted crossing severe stenoses, and the ability to capture small particles and soluble vasoactive molecules such as endothelin and serotonin as well as coagulation components that have been shown to be liberated during SVG PCI.¹⁴ Disadvantages included the need to completely occlude the vein graft for several minutes, which was not well tolerated by some patients, and the requirement for several disease-free centimeters beyond the lesion to place the occlusion balloon. The complexity of this form of embolic protection compared to filters led to underutilization, and it was eventually removed from the market.

The effectiveness of filter-based embolic protection during SVG stent implantation was evaluated in the FilterWire EX During Transluminal Intervention of Saphenous Vein Grafts (FIRE) trial, a randomized study of 650 patients in 65 centers.¹⁵ Thirty-day MACE rate was 9.9% with the filter-based EPD and 11.6% with PercuSurge with virtually identical rates of MI (9% vs 10%) and death (0.9% vs 0.9%). Advantages of filter-based EPD included ease of use, maintenance of flow, avoidance of ischemia during stenting, and good visualization. Disadvantages included the need to cross the lesion with a somewhat bulky filter device that might require predilation, which has been shown to increase MI; the need for a substantial “landing zone” beyond the lesion; and the inability to remove soluble factors and particles smaller than the filter pores (<100 μm) (Fig. 36-1). The latter factors may not be clinically important based on comparisons such as FIRE. In addition, there can be difficulty with filter withdrawal due to inability to pass the retrieval catheter through a newly placed stent. This is especially true when the stent is in an aorto-ostial location with stent struts protruding into the aorta. Guide catheter–induced distortion of the newly placed ostial stent or entrapment of the filter on the distal stent edge can lead to major complications including stent embolization and thrombosis.¹⁶ Finally, filters have finite limits as to the amount of debris that can be captured with maintenance of flow. Large atheroembolic or thrombotic loads can result in filter occlusion and the appearance of no reflow (Figs. 36-2 and 36-3). The optimal strategies, should this occur, include aspiration of the stagnant “dye” column, which frequently contains suspended particles; removal of the filter; and placing another filter if more interventional work is required. A variety of filters are currently available and appear to be equally effective.

A third form of EPD, known as PROXIS, which used a proximal occluding balloon, was tested in a randomized trial and was comparable to either filter or Guardwire.¹⁷ Advantages of proximal occlusion included protection during wire passage, side branch protection, and no need for a distal parking segment, a problem in many patients that prevented use of filters and PercuSurge. Unfortunately, as with the Guardwire, the complexity of proximal occlusion-aspiration and the requirement for graft occlusion for several minutes resulted in little usage and withdrawal of the device from the market.

The early report of Liu et al¹⁰ showing that plaque volume was a predictor of MI in SVG PCI was extended by Coolong and colleagues.¹⁸ They showed in about 4000 patients that degeneration score and plaque volume were the most important predictors of MI, but the presence of thrombus, advanced age, and active smoking also contributed.¹⁸ When the occurrence of MACE was analyzed relative to these predictors, embolic protection was beneficial across all risk groups, implying that embolic protection should be used routinely in all suitable patients undergoing PCI in old SVGs as recommended in the 2011 PCI guidelines.¹⁹ However, data from the National Cardiovascular Date Registry indicated that embolic protection was used in only 21% of 49,325 senior patients who underwent SVG PCI and that acute and 3-year outcomes were not better in propensity score–matched patients in whom EPDs were used.²⁰ In addition, in the recent ISAR-CABG (Is Drug-Eluting Stenting Associated With Improved Results in Coronary Artery Bypass Grafts?) study in which 610 patients underwent stent implantation in SVGs, the occurrence of MACE at 30 days was only 4% despite the use of embolic protection in <5% of patients.²¹ These observations cause one to question whether practices in place at the time of SAFER may have led to an increased rate of MI in the control group. For example, 40% of control patients underwent postdeployment stent dilation, a practice currently avoided by most experienced operators because further stent expansion frequently leads to no reflow. In addition, postdilation was carried out with a balloon that had a mean diameter of 4.2 mm, whereas the mean reference vessel diameter was only 3.4 mm. Use of a balloon larger than the reference vessel increased MI in the study by Iakovau et al.⁷ In addition only 27% of patients in the Guardwire arm underwent postdilation of the stent. These practice patterns may have contributed to the higher rate of periprocedural MI in the SAFER control group.

Debulking procedures in SVGs using directional atherectomy, the transluminal extraction catheter, and the excimer laser have been shown to have increased complexity and costs without evidence of improved outcomes. Consequently, atheroablative strategies do not play a significant role in current percutaneous intervention in venous or arterial grafts. Mechanical thrombectomy is currently feasible with a variety of techniques, including aspiration with guide catheter, guide catheter extenders, or mono-rail aspiration catheters or by Angiojet thrombectomy (Boston Scientific, Marlborough, MA) (Fig. 36-4). Although proof of benefit is lacking, anecdotal experiences have convinced most practitioners that proceeding with PCI in the face of a large thrombus burden is rarely advisable when mechanical thrombectomy or some other strategy to eliminate or reduce thrombus is feasible (see later “Large Thrombus Burden” section).

Observational studies of the use of drug-eluting stents (DESs) in SVGs suggested that they were safe and relatively effective. A multicenter registry of over 1000 patients who underwent SVG PCI reported that, at 9 months, patients receiving DES had significantly lower rates of death or MI (hazard ratio [HR], 0.52) and target vessel revascularization (HR, 0.36) compared to patients treated with bare metal stents (BMSs), and at 2 years, the mortality rate was lower (HR, 0.60).²² Multiple meta-analyses comparing DES with BMS in SVG have been published. In a meta-analysis, Mamas and colleagues²³ analyzed 20 studies with over 5000 patients and reported no safety concerns and a 36% decrease in MACE in DES-treated patients.²³ Three randomized trials have been published comparing sirolimus-eluting stents (SESs) or paclitaxel-eluting stents (PESs) with BMS in SVGs.^21,24-27 Vermeersch et al^24,25 randomized 75 patients to SES or BMS and found reduced restenosis at 6 months with SES, but at 3 years, target vessel revascularization was similar, suggesting a “catch-up” phenomenon, and SES-treated patients had a higher mortality. In the second randomized trial, 80 patients were randomly assigned to receive PES or BMS, and at 1 year, outcomes were more favorable with PES with respect to restenosis (11% vs 57%; P < .0001) and target lesion revascularization (5% vs 28%; P = .003), and clinical outcomes were better in PES-treated patients at 3 years.^26,27

The third randomized trial, ISAR-CABG, was the only study powered to compare clinical outcomes.²¹ Six-hundred ten patients were enrolled and randomly assigned to receive a DES (sirolimus or paclitaxel eluting) or BMS. Inclusion criteria were broad, and clinical characteristics of the 2 groups were similar. Outcomes at 1 year are reported in Table 36-1. The incidence of MACE, the primary end point, was 15% with DES and 22% with BMS, a 36% risk reduction. Target lesion and target vessel revascularization rates were significantly lower in DES-treated patients. There were no significant differences in all-cause mortality, MI, or definite or probable stent thrombosis. The 36% reduction in MACE is the same reported in a meta-analysis²³ and in the stent registry.²² It is important to note that follow-up in ISAR-CABG was only 1 year and that longer term outcomes are needed.

Three recent publications of large registries reported somewhat differing long-term results. In a Danish registry, there was no benefit of DES with respect to death or stent failure at 3 years.²⁸ However, in a large cohort of 709 propensity score–matched pairs of patients, the need for target vessel revascularization at 4 years was significantly less in DES-treated patients (21% vs 28%; P = .004; a 24% risk reduction).²⁹ In patients with diabetes or stented segments ≥30 mm, the number of procedures needed to prevent a target vessel revascularization was 8 and 7, respectively. In a report of stent implantation in SVGs from 63 Veterans Affairs hospitals for a 4-year period ending in 2011, DES use was associated with lower mortality in propensity-matched patients.³⁰ Whether some late erosion of benefit with DES in SVG is related to the well-described delay in neointimal hyperplasia observed with DES in SVG is uncertain but plausible.

Virtually all the data available regarding long-term pathologic and clinical studies of DES in SVGs involve first-generation SESs and PESs. However, second-generation stents, which have been shown to be superior in native coronary arteries, are currently being implanted in SVGs. In an observational single-center study, 2-year outcomes of patients receiving second-generation everolimus-eluting stents (EESs; n = 88) were shown to be superior to those of patients receiving first-generation SESs or PESs (n = 243) with respect to MACE (18% vs 35%; P = .003), target lesion revascularization (1% vs 12%; P = .005), and target vessel revascularization (7% vs 25%; P < .001).³¹ However, patients in the 2 groups were treated at different time periods and subject to selection bias. Although the use of DES in SVGs is somewhat controversial,³² given the paucity of long-term clinical outcome data and cost-effectiveness of current stents in SVGs, they do appear to be safe and confer at least short-term benefit, and their use is supported by the 2011 PCI guidelines.¹⁹ Patients most likely to benefit are those with diabetes, long lesions, SVGs <3.5 mm in diameter, and the ability to comply with long-term dual antiplatelet therapy.

The results of randomized trials comparing polytetrafluoroethylene (PTFE)-covered stents with BMSs in SVGs were not encouraging.^33,34 Six-month MACE rates were higher in patients receiving covered stents due to more MIs and late occlusion. However, PTFE-covered stents remain an important, potentially lifesaving adjunct for treatment of patients with SVG or left internal mammary artery (LIMA) graft perforation.