American College of Cardiology

The outcomes of coronary interventional procedures are measured in terms of success and complications and are related to the mechanisms of the employed devices, as well as the clinical and anatomic patient-related factors. Complications can be divided into 2 categories: (1) those common to all arterial catheterization procedures and (2) those related to the specific technology used for the coronary procedure. Specific definitions of success and complications exist, and where appropriate, the definitions used herein are consistent with the ACC-National Cardiovascular Data Registry™ Catheterization Laboratory Module Version 2.0 ⁽¹⁵⁾. With increased operator experience, new technology, and adjunctive pharmacotherapy, the overall success and complication rates of angioplasty have improved.

A. Definitions of PCI Success

The success of a PCI procedure may be defined by angiographic, procedural, and clinical criteria.

1. Angiographic Success. A successful PCI produces substantial enlargement of the lumen at the target site. The consensus definition prior to the widespread use of stents was the achievement of a minimum stenosis diameter reduction to <50% in the presence of grade 3 TIMI flow (assessed by angiography) ⁽¹⁶⁾. However, with the advent of advanced adjunct technology, including coronary stents, a minimum stenosis diameter reduction to <20% has been the clinical benchmark of an optimal angiographic result. Frequently, there is a disparity between the visual assessment and computer-aided quantitative stenosis measurement ^{(17, 18)}, and the determination of success may be problematic when success rates are self-reported.

2. Procedural Success. A successful PCI should achieve angiographic success without in-hospital major clinical complications (e.g., death, MI, emergency coronary artery bypass surgery) during hospitalization ^(2,16). Although the occurrence of emergency coronary artery bypass surgery and death are easily identified endpoints, the definition of procedure-related MI has been debated. The development of Q-waves in addition to a threshold value of CK elevation has been commonly used. However, the significance of enzyme elevations in the absence of Q-waves remains a subject of investigation and debate. Several reports have identified non-Q-wave MIs with CK-MB elevations 3-5 times the upper limit of normal as having clinical significance ^(19,20). Thus a significant increase in CK-MB without Q-waves is considered by most to qualify as an associated complication of PCI.

3. Clinical Success. In the short term, a clinically successful PCI includes anatomic and procedural success with relief of signs and/or symptoms of myocardial ischemia after the patient recovers from the procedure. The long-term clinical success requires that the short-term clinical success remains durable and that the patient has persistent relief of signs and symptoms of myocardial ischemia for more than 6 months after the procedure. Restenosis is the principal cause of lack of long-term clinical success when a short-term clinical success has been achieved. Restenosis is not considered a complication but rather an associated response to vascular injury. The frequency of clinically important restenosis may be judged by the frequency with which subsequent revascularization procedures are performed on target vessels after the index procedure. A very high rate of restenosis may suggest that the operator chooses an excess of lesions which are likely to restenose, such as long lesions or those involving small vessels.

B. Definitions of Procedural Complications

As outlined in the 1998 coronary interventional document ⁽²¹⁾, procedural complications are divided into 6 basic categories: death, MI, emergency coronary artery bypass graft (CABG), stroke, vascular access site complications, and contrast agent nephropathy. Key data elements and definitions to measure the clinical management and outcomes of patients undergoing diagnostic catheterization and/or PCI have been defined in the Clinical Data Standards document ⁽²²⁾ and the ACC-National Cardiovascular Data Registry™ Catheterization Laboratory Module version 2.0 ⁽¹⁵⁾. These rigorous definitions for key adverse events are endorsed by this Writing Committee for inclusion in the present PCI Guidelines (Table 1).

Notably, the definition of MI has evolved over the past several years. It should be emphasized that the simple categorization of MI into 2 classes based on the development of new Q-waves alone is no longer sufficient as a classification scheme for measuring MI following PCI. Since the measurement of CK and CK-MB are widely available, myocardial necrosis may be measured with a high level of sensitivity and specificity, regardless of the clinical presentation and associated ECG findings. The use of CK-MB for measuring myocardial necrosis is preferable to a less sensitive and less specific CK determination. The mass determination of CK-MB is now commonly used at most hospitals, and elevations of this myocardial specific enzyme are reported in nanograms per deciliter. Cardiac troponin T and I have now been introduced as measurements of myocardial necrosis and have been proven to be more sensitive and specific than CK-MB. However, prognostic criteria after PCI based on troponin T and I have not yet been developed.

Since normal values may vary among hospitals and selected patient subsets, an index of the measured value is usually reported in terms of the value of the upper limit of normal (i.e., CK-MB index of 3 corresponds to an elevation of CK-MB to 3 times its upper limit of normal value). Thus, myocardial necrosis may be determined as an abnormally elevated CK-MB index (>1), based upon 2 or 3 serial determinations during the 18 to 24 h after coronary intervention and the abnormality may range from a low index (1 to 3 times normal) with no or non-specific ECG findings, to a high index (>10 to 15 times normal) with significant ECG findings including the development of new Q-waves.

If serial determinations are performed after PCI, an abnormally high value (CK-MB>1 times normal) can be expected in 10 to 15% of balloon angioplasty procedures, 15 to 20% of stent procedures, 25 to 35% of atherectomy procedures, and >25% for any device used in saphenous vein grafts (SVGs) or long lesions with a high atherosclerotic burden, even in the absence of other signs and symptoms of MI. There is no accepted consensus on what level of CK-MB index (with or without clinical or electrocardiographic [ECG] findings) is indicative of a clinically important MI following the interventional procedure. The Writing Committee recommends that a CK-MB determination be performed on all patients who have signs or symptoms of suggestive MI following the procedure or in patients in whom there is angiographic evidence of abrupt vessel closure, important side branch occlusion, or new and persistent slow coronary flow. In patients in whom a clinically driven CK-MB determination is made, a CK-MB of >3 times the upper limit of normal would constitute a clinically significant MI. These relationships may be confounded by other factors, such as atherosclerosis.

C. Acute Outcome

Despite the extension of coronary intervention to higher-risk patients with comorbid disease and complex coronary anatomy, angiographic and procedural success have increased since the first National Heart Lung and Blood Institute (NHLBI) registries with an associated decrease in the major complications of Q-wave MI and emergency CABG (Table 2) ^(2,6,23,24). Improvements in balloon technology coupled with the increased use of non-balloon devices, particularly stents (which are effective in treating abrupt vessel closure) ⁽²⁵⁾ and glycoprotein IIb/IIIa platelet receptor antagonists ^(26-28) have favorably influenced acute procedural outcome. This combined balloon/device/ pharmacologic approach to coronary intervention in elective procedures has resulted in angiographic success rates of 96 to 99%, with Q-wave MI rates of 1 to 3%, emergency coronary artery bypass surgery rates of 0.2 to 3%, and unadjusted inhospital mortality rates of 0.5 to 1.4% ^(29-34). The integrated approach utilizing adjunct pharmacologic therapies and the enormous increase in the use of stents as a primary strategy have resulted in an improved procedural outcome of balloon angioplasty ⁽³⁵⁾. Improved balloon/pharmacologic techniques may achieve results comparable to those obtained with stents with the ability to perform provisional (for suboptimal result) or bail-out (for acute or threatened vessel closure) stent deployment.

It should be noted, however, that the incidence of elevated creatine kinase has increased in the new device era ⁽³⁶⁾. The significance of this finding, in the absence of a clinical event, is uncertain and the subject of ongoing debate. This issue is discussed in more detail in Section VI, C.1. Post-Procedure Evaluation of Ischemia.

D. Long-Term Outcome and Restenosis

Although improvements in technology, including stents and new pharmacologic therapy, have resulted in an improved acute outcome of the procedure, the impact of these changes on long-term (5-10 years) outcome may be less dramatic where factors such as advanced age, reduced left ventricular (LV) function, and complex multivessel disease in patients currently undergoing PCI may have a more important influence. In addition, available data on long-term outcome are mostly limited to patients undergoing PTCA. 10-Year follow-up of the initial cohort of patients treated with PTCA revealed an 89.5% survival rate (95% in patients with single-vessel disease, 81% in patients with multivessel disease) ⁽⁴⁵⁾. In patients undergoing within the 1985-1986 NHLBI PTCA Registry ⁽⁴⁶⁾, 5-year survival was 92.9% for patients with single-vessel disease, 88.5% for those with 2-vessel disease, and 86.5% for those with 3-vessel disease. In patients with multivessel disease undergoing PTCA in BARI ⁽⁹⁾, 5-year survival was 86.3%, and infarct-free survival was 78.7%. Specifically, 5-year survival was 84.7% in patients with 3-vessel disease and 87.6% in patients with 2-vessel disease.

In addition to the presence of multivessel disease, other clinical factors adversely impact late mortality. In randomized patients with treated diabetes in BARI, the 5-year survival was 65.5%, and the cardiac mortality was 20.6% in comparison to 5.8% cardiac mortality in patients without treated diabetes ⁽⁴⁷⁾, although among eligible but not randomized diabetic patients, the 5-year cardiac mortality was 7.5% ⁽⁴⁸⁾. In the 1985-1986 NHLBI PTCA Registry, 4-year survival was significantly lower in women (89.2%) in comparison to men (93.4%) ⁽⁴⁹⁾. In addition, although LV dysfunction was not associated with an increase in in-hospital mortality or nonfatal MI in patients undergoing PTCA in the same registry, it was an independent predictor of a higher long-term mortality ⁽⁵⁰⁾.

A major determinant of event-free survival following coronary intervention is the incidence of restenosis which had, until the development of stents, remained fairly constant, despite multiple pharmacologic and mechanical approaches to limit this process (Table 3). Depending on the definition, (i.e., whether clinical or angiographic restenosis or target lesion revascularization is measured), the incidence of restenosis following coronary intervention had been 30 to 40%, and higher in certain clinical and angiographic subsets ⁽⁵¹⁾.

The pathogenesis of the response to mechanical coronary injury is thought to relate to a combination of growth factor stimulation, smooth muscle cell migration and proliferation, organization of thrombus, platelet deposition, and elastic recoil ^(69,70). In addition, dynamic change in vessel size (or lack of compensatory enlargement) has been implicated ⁽⁷¹⁾. It has been suggested that attempts to reduce restenosis have failed, in part due to lack of recognition of the importance of this factor ⁽⁷²⁾. Although numerous definitions of restenosis have been proposed, >50% diameter stenosis at follow-up angiography has been most frequently used. However, it is now recognized that the response to arterial injury is a continuous rather than a dichotomous process, occurring to some degree in all patients ⁽⁷³⁾. Therefore, cumulative frequency distributions of the continuous variables of minimal lumen diameter or percent diameter stenosis are now used to evaluate restenosis in large patient populations ⁽⁷⁴⁾ (Figure 2).

Although multiple clinical factors (diabetes, unstable angina, acute MI, prior restenosis) ^(75,76), angiographic factors (proximal left anterior descending artery, small vessel diameters, total occlusion, long lesion length, SVG) ⁽⁷⁷⁾, and procedural factors (higher post-procedure percent diameter stenosis, smaller minimal lumen diameter, and smaller acute gain) ⁽⁷⁴⁾ have been associated with an increased incidence of restenosis, the ability to integrate these factors and predict the risk of restenosis in individual patients following the procedure remains difficult. The most promising potential approaches to favorably impact the restenosis process relate to (1) the ability to decrease elastic recoil and remodeling using intracoronary stents, and (2) to the ability to reduce intimal hyperplasia using catheter-based ionizing radiation. More than 6300 patients have been studied in 12 randomized clinical trials to assess the efficacy of PTCA versus stents to reduce restenosis (Table 4).

The pivotal BENESTENT ⁽³²⁾ and STRESS Trials ⁽³¹⁾ documented that stents significantly reduce angiographic restenosis in comparison to balloon angioplasty (BENESTENT: 22% vs. 32%; STRESS: 32% vs. 42% respectively). These results have been corroborated in the BENESTENT II trial in which the angiographic restenosis rate was reduced by 45% (from 31 to 16% in patients treated with balloon angioplasty versus heparin-coated stents, respectively) ⁽⁶⁶⁾.

In addition, randomized studies in patients with in-stent restenosis have shown that both intracoronary gamma and beta radiation significantly reduced the rate of subsequent angiographic and clinical restenosis by 30 to 50% ^(78-81). Late subacute thrombosis was observed in some of these series ⁽⁸²⁾, but this syndrome has resolved with judicious use of stents and extended adjunct antiplatelet therapy with ticlopidine or clopidogrel. Also, in a preliminary study of patients undergoing successful balloon angioplasty, delivery of intracoronary beta radiation resulted in a restenosis rate of 15% ⁽⁸³⁾.

When technically feasible, in patients who experience restenosis, it is standard practice to perform repeat PCI. In this setting, stents are being used with the hope of decreasing the rate of subsequent restenosis. However, in-stent restenosis, particularly when diffuse, represents a challenging problem. The efficacy of various treatment modalities for in-stent restenosis is under active investigation.

E. Predictors of Success/Complications

1. Anatomic Factors. Target lesion anatomic factors related to adverse outcomes have been widely examined. Lesion morphology and absolute stenosis severity were identified as the prominent predictors of immediate outcome during PTCA in the pre-stent era ^(93,94). Abrupt vessel closure, due primarily to thrombus or dissection, was reported in 3 to 8% of patients and was associated with certain lesion characteristics ^(95-97). The risk of PTCA in the pre-stent era relative to anatomic subsets has been identified in previous NHLBI PTCA Registry data ⁽⁶⁾ and by the ACC/AHA Task Force ^(16,98). The lesion classification based on severity of characteristics proposed in the past ^(98-100) has been principally altered using the present PCI techniques which capitalize on the ability of stents to manage initial and subsequent complications of coronary interventions ⁽¹⁰¹⁾. As a result the Committee has revised the previous ACC/AHA lesion classification system to reflect low, moderate, and high risk (Table 5) in accordance with the PCI Clinical Data Standards from the ACC-National Cardiovascular Data Registry™ ⁽¹⁵⁾.

2. Clinical Factors. Coexistent clinical conditions can increase the complication rates for any given anatomic risk factor. For example, complications occurred in 15.4% of diabetic patients vs. 5.8% of nondiabetic patients undergoing balloon angioplasty in a multicenter experience ^(94,97). Several studies have reported specific factors associated with increased risk of adverse outcome following balloon angioplasty. These factors include advanced age, female gender, unstable angina, congestive heart failure (CHF), diabetes, and multivessel CAD ^{(9,93,94,102,103)} (Table 6). The BARI trial found that patients with diabetes and multivessel CAD had an increased periprocedural risk of ischemic complications and increased 5-year mortality in comparison to patients without diabetes or in comparison to patients with diabetes undergoing bypass surgery using internal thoracic arterial grafts ^(9,38). Patients with impaired renal function, especially diabetics, are at increased risk for contrast nephropathy ⁽¹⁰⁴⁾ and increased 30-day and 1-year mortality.

Increased risk for severe compromise in LV function or fatal outcome may occur with a complication of a vessel that also supplies collateral flow to viable myocardium. Certain variables were used to prospectively identify patients at risk for significant cardiovascular compromise during PTCA ^(105,106). These resulted in a composite 4-variable scoring system, prospectively validated to be both sensitive and specific in predicting cardiovascular collapse for failed PTCA and includes (1) percentage of myocardium at risk (e.g., >50% viable myocardium at risk and LV ejection fraction of <25%), (2) pre-angioplasty percent diameter stenosis, (3) multivessel CAD, and (4) diffuse disease in the dilated segment ⁽¹⁰⁷⁾ or a high myocardial jeopardy score ⁽¹⁰⁸⁾. Patients with higher pre-procedural jeopardy scores were shown to have a greater likelihood of cardiovascular collapse when abrupt vessel closure occurred during PTCA ⁽¹⁰⁵⁾. The clinical risk factors associated with inhospital adverse events have been further evaluated with additional experience during the PCI era and summarized based on odds ratio >2.0 or results of multivariate analysis (Table 6).

3. Risk of Death. In the majority of patients undergoing elective PCI, death as a result of PCI is directly related to the occurrence of coronary artery occlusion and is most frequently associated with pronounced LV failure ^(105,106) (Table 6). The clinical and angiographic variables associated with increased mortality include advanced age, female gender, diabetes, prior MI, multivessel disease, left main or equivalent coronary disease, a large area of myocardium at risk, pre-existing impairment of LV or renal function, and collateral vessels supplying significant areas of myocardium that originate distal to the segment to be dilated (Table 6) ^{(9,93,95,97,102-105,107-110)}.

4. Women. In comparison to men, women undergoing PCI are older and have a higher incidence of hypertension, diabetes mellitus, hypercholesterolemia, and comorbid disease ^(49,111-114). Women also have more unstable angina and a higher functional class of stable angina (Canadian Cardiovascular Society Class III and IV) for a given extent of disease ⁽¹¹⁵⁾. Yet, despite the higher-risk profile in women, the extent of epicardial coronary disease is similar (or less) in comparison to men. In addition, although women presenting for revascularization have less multivessel disease and better LV systolic function, the incidence of CHF is higher in women than in men. The reason for this gender paradox is unclear, but it has been postulated that women have more diastolic dysfunction, perhaps based on older age and hypertension, in comparison to men ⁽¹¹⁶⁾.

Early reports of patients undergoing PTCA revealed a lower procedural success rate in women ⁽¹¹²⁾; however, more recent studies have noted similar angiographic outcome and incidence of MI and emergency coronary artery bypass surgery in women and men ⁽⁴⁹⁾. Although reports have been inconsistent, in several large-scale registries, in-hospital mortality is significantly higher in women, and an independent effect of gender on acute mortality following PTCA persists after adjustments for the baseline higher-risk profile in women ^(49,117). The reason for the increase in mortality is unknown, but small vessel size and hypertensive heart disease in women have been thought to play a role. Although a few studies have noted that gender is not an independent predictor of mortality when body surface area (a surrogate for vessel size) is accounted for ⁽¹¹¹⁾, the impact of body size on outcome has not been thoroughly evaluated. The higher incidence of vascular complications, coronary dissection, and perforation in women undergoing coronary intervention has been attributed to the smaller vasculature in women in comparison to men. In addition, diagnostic intravascular ultrasound (IVUS) studies have not detected any gender-specific differences in plaque morphology or luminal dimensions once differences in body surface area were corrected, suggesting that differences in vessel size account for some of the apparent early and late outcome differences previously noted in women ⁽¹¹⁸⁾. It has also been postulated that the volume shifts and periods of transient ischemia during coronary angioplasty are less well tolerated by the hypertrophied ventricle in women, and CHF has shown to be an independent predictor of mortality in both women and men undergoing coronary angioplasty ⁽¹¹⁹⁾.

An improved outcome has been reported in women undergoing both coronary balloon and new device angioplasty, despite the fact that the women (similar to men) are older and with more complex disease than women treated previously. In fact, in the 1993-1994 NHLBI PTCA Registry (open to women only), procedural success was higher and major complications lower in comparison to women treated in the 1985-1986 registry ⁽²⁴⁾. Additionally, patients undergoing balloon angioplasty in BARI, in-hospital mortality, MI, emergency coronary artery bypass surgery rates, and 5-year mortality were similar in women and men, although women had a higher incidence of periprocedural CHF and pulmonary edema ⁽¹²⁰⁾.

In a registry of 373 consecutive patients undergoing directional coronary atherectomy (DCA), although early and late outcomes were similar, the lower procedural success observed in women (73% vs. 83%, p = 0.011) was again attributed to their smaller vessel caliber ⁽¹²¹⁾. Therefore, although women presenting for coronary revascularization have a higher-risk profile, currently the acute and long-term outcomes are similar to those in men. Much of the increase in adverse outcome seen in women can be accounted for by comorbidities, although gender imparts a small independent effect. Finally, it is important to note that in women undergoing coronary intervention, the acute outcome has improved and the long-term outcome remains excellent. Therefore, coronary intervention should be considered for women in need of revascularization with the anticipation of a favorable outcome (Table 7).

5. The Elderly Patient. Age >75 years is one of the major clinical variables associated with increased risk of complications ⁽¹²⁵⁾. In the elderly population, the morphologic and clinical variables are compounded by advanced years with the very elderly having the highest risk of adverse outcomes ⁽¹²⁶⁾. In octogenarians, although feasibility has been established for most interventional procedures, the risk of both percutaneous and nonpercutaneous revascularization is increased ^(127,128). Octogenarians undergoing percutaneous intervention have a higher incidence of prior MI, lower LV ejection fraction, and more frequent CHF ⁽¹²⁹⁾. In the stent era, procedural success rates and short-term outcomes are comparable to those for nonoctogenarians ⁽¹³⁰⁾. Thus, with rare exception (primary PCI for cardiogenic shock for patients >75 years), a separate category has not been created in these Guidelines for the elderly. However, their higher incidence of comorbidities should be taken into account when considering the need for PCI.

6. Diabetes Mellitus. In the TIMI-IIB study of myocardial infarction, patients with diabetes mellitus had significantly higher 6-week (11.6% vs. 4.7%), 1-year (18.0% vs. 6.7%), and 3-year (21.6% vs. 9.6%) mortality rates compared to nondiabetic patients ⁽¹³¹⁾. Patients with diabetes with a first MI who were randomly assigned to the early invasive strategy faired worse than those managed conservatively (42-day mortality: death or MI, or death alone 14.8% vs. 4.2%; p < 0.001) ⁽¹³²⁾. Early catheterization and intervention strategy after thrombolysis was of little benefit in these patients with diabetes. Routine catheterization and angioplasty in this patient subgroup should be based on clinical need and ischemic risk stratification.

Stenting decreases the need for target revascularization procedures in diabetic patients compared with balloon angioplasty. The efficacy of stenting with glycoprotein IIb/IIIa inhibitors was assessed in the diabetic population compared to those without diabetes in a substudy of the EPISTENT trial ⁽¹³³⁾. One hundred seventy-three diabetic patients were randomized to stent/placebo combination, 162 patients to stent/abciximab combination, and 156 patients to balloon angioplasty/abciximab combination. For the composite endpoint of death, MI, or target vessel revascularization, the rates were as follows: 25%, 23%, and 13% for the stent/placebo, balloon/abciximab, and stent/abciximab groups (p = 0.005). Irrespective of revascularization strategy abciximab significantly reduced 6-month death and MI rate in patients with diabetes for all strategies. Likewise, 6-month target vessel revascularization was reduced in the stent/abciximab group approach. One-year mortality for diabetics was 4.1% for the stent/placebo group and 1.2% for the stent/abciximab group. Although this difference was not significant, the combination of stenting and abciximab among diabetics resulted in a significant reduction in 6-month rates of death and target-vessel revascularization compared to stent/placebo or balloon angioplasty/abciximab therapy ⁽¹³³⁾. The BARI trial, in which stents and abciximab were not used, showed that survival was better for patients with treated diabetes undergoing CABG with an arterial conduit than for those undergoing angioplasty. A discussion about the selection of diabetic patients for surgical revascularization or PCI may be found in Section III. Outcomes, F. Comparison With Bypass Surgery.

7. Coronary Angioplasty After Coronary Artery Bypass Surgery. Although speculated to be at higher risk, patients having PCI of native vessels after prior coronary bypass surgery have, in recent years, nearly equivalent interventional outcomes and complication rates compared to patients having similar interventions without prior surgery. For PCI of SVG, studies indicate that the rate of successful angioplasty exceeds 90%, death <1.2%, and Q-wave MI <2.5% (Table 8). The incidence of non-Q-wave MI may be higher than that associated with native coronary arteries ^(134-136).

In consideration of PCI for SVG, the age of the SVG and duration and severity of myocardial ischemia should be taken into consideration. Use of GP IIb/IIIa blockers has not been shown to improve results of angioplasty in vein grafts. The native vessels should be treated with PCI if feasible. Patients with older and/or severely diseased SVGs may benefit from elective repeat coronary artery bypass graft surgery rather than PCI ^(137,138).

In some circumstances, PCI of a protected left main coronary artery stenosis with a patent and functional left anterior descending or left circumflex coronary conduit can be considered. PCI should be recognized as a palliative procedure with the potential to delay the ultimate application of repeat CABG surgery.

8. Specific Technical Considerations. Certain outcomes of PCI may be specifically related to the technology utilized for coronary recanalization. The occurrence of periprocedural CPK-MB elevation 3 times the upper limit of normal appears to occur more frequently following use of ablative technology such as rotational or directional atherectomy ^{(20,34,58,140,146)}. Antecedent unstable angina appears to be a clinical predictor of slow flow and periprocedural infarction following ablative technologies ⁽¹⁴⁷⁾, and direct platelet activation has been demonstrated to occur with both directional and rotational atherectomy ⁽¹⁴⁸⁾. In support of the premise that platelets play a pathophysiologic role in periprocedural MI are observations that the presence and magnitude of CK-MB elevation following ablative technologies can be reduced to levels observed following balloon angioplasty by the administration of prophylactic platelet GP IIb/IIIa receptor blockade ^(149,150).

Coronary perforation may occur more commonly following the use of ablative technologies, including rotational, directional or extraction atherectomy, and excimer laser coronary angioplasty. However, the incidence of perforation has been reported variably to be 0.10 to 1.14% with balloon angioplasty; 0.25 to 0.70% with directional coronary atherectomy; 0.0 to 1.3% with rotational atherectomy; 1.3 to 2.1% with extraction atherectomy; and, 1.9 to 2.0% following excimer laser coronary angioplasty ^(151,152). Coronary perforation complicates PCI more frequently in the elderly and in women. While 20% of perforations may be secondary to the coronary guidewire, most are related to the specific technology used. Perforation is usually (80 to 90%) evident at the time of the interventional procedure and should be a primary consideration in the differential diagnosis for cardiac tamponade manifest within 24 h of the procedure. Perforations may be classified based on angiographic appearance as Type I—extraluminal crater without extravasation; Type II—pericardial and myocardial blush without contrast jet extravasation; and Type III—extravasation through a frank (1 mm) perforation ⁽¹⁵¹⁾. In the absence of extravasation (Type III), the majority of perforations may be effectively managed without urgent surgical intervention. Type III perforations have been successfully managed nonoperatively with pericardiocentesis, reversal of anticoagulation, and either prolonged perfusion balloon inflation at the site of perforation or deployment of a covered stent. If these approaches are not successful, perforations caused by directional atherectomy catheters usually require surgical repair (Table 9).

9. Issues of Hemodynamic Support in High-Risk Angioplasty. Controversy exists about the ability to predict hemodynamic compromise during coronary angioplasty. Hemodynamic compromise, defined as a decrease in systolic blood pressure to an absolute level <90 mm Hg during balloon inflation, was often associated with LV ejection fraction <35%, >50% of myocardium at risk, and PTCA performed on the last remaining vessel ^(95,107).

Early feasibility studies of high-risk PTCA using percutaneous cardiopulmonary support (CPS) indicated that although initial likelihood of success was high, vascular morbidity was also high with an incidence of 43% ^(153,154). However, no study has published data to validate commonly employed high-risk categorization.

Elective high-risk PCI can be performed safely without intra-aortic balloon pump (IABP) or CPS in most circumstances. Emergency high-risk PCI such as direct PCI for acute MI can usually be performed without IABP or CPS. CPS for high-risk PCI should be reserved only for patients at the extreme end of the spectrum of hemodynamic compromise, such as those patients with extremely depressed LV function and patients in cardiogenic shock. However, it should be noted that in patients with borderline hemodynamics, ongoing ischemia, or cardiogenic shock, insertion of an intra-aortic balloon just prior to coronary instrumentation has been associated with improved outcomes ^(155,156). Furthermore, it is reasonable to obtain vascular access in the contralateral femoral artery prior to the procedure in patients in whom the risk of hemodynamic compromise is high, thereby facilitating intra-aortic balloon insertion, if necessary.

For high-risk patients, clinical and anatomic variables influencing complications and outcome should be assessed before the performance of PCI to determine procedural risk, the risk of abrupt vessel closure, and potential for cardiovascular collapse. In patients having a higher-risk profile, consideration of alternative therapies, particularly coronary bypass surgery, formalized surgical standby, or periprocedural hemodynamic support should be addressed before proceeding with PCI.

F. Comparison With Bypass Surgery

The major advantage of PCI is its relative ease of use, avoiding general anesthesia, thoracotomy, extracorporeal circulation, CNS complications, and prolonged convalescence. Repeat PCI can be performed more easily than repeat bypass surgery, and revascularization can be achieved more quickly in emergency situations. The disadvantages of PCI are early restenosis and the inability to relieve many totally occluded arteries and/or those vessels with extensive atherosclerotic disease.

Coronary artery bypass surgery has the advantages of greater durability (graft patency rates exceeding 90% at 10 years with arterial conduits) ⁽¹⁵⁷⁾ and more complete revascularization irrespective of the morphology of the obstructing atherosclerotic lesion. Generally speaking, the greater the extent of coronary atherosclerosis and its diffuseness, the more compelling the choice of coronary artery bypass surgery, particularly if LV function is depressed. Patients with lesser extent of disease and localized lesions are good candidates for endovascular approaches.

PTCA and coronary artery bypass surgery have been compared in many nonrandomized and randomized studies. The most accurate comparisons of outcomes are best made from prospective randomized trials of patients suitable for either treatment. Although results of these trials provide useful information for selection of therapy in several patient subgroups, prior studies of PTCA may not reflect outcome of current PCI practice, which includes frequent use of stents and antiplatelet drugs. Similarly, many previous studies of CABG may not reflect outcome of current surgical practice in which arterial conduits are used whenever practicable. Beating heart bypass operations are also employed for selected patients with single-vessel disease with reduced morbidity ⁽¹⁵⁸⁾. In addition, patients are selected for PCI (with or without stenting) because of certain lesion characteristics, and these anatomical criteria are not required for CABG.

Randomized trials also must be interpreted carefully. It is unethical to withhold subsequent PCI or CABG from patients solely because they fail an earlier treatment; thus, comparative prospective studies can only compare initial strategies of revascularization. This critically important point is frequently overlooked by those who claim that a randomized study proves equally good outcome of one method of revascularization over the other. Indeed, it would seem highly unlikely that any randomized trial of PCI and CABG could demonstrate a survival advantage of an initial revascularization method as long as frequent crossover to alternate and/or new therapies is allowed.

Despite these limitations, some generalizations can be made from comparative trials of PTCA and CABG. First, for most patients with single-vessel disease, late survival is similar with either revascularization strategy, and this might be expected given the generally good prognosis of most patients with single-vessel disease managed medically ^(159-161).

Two prospective clinical trials have evaluated PTCA and CABG for revascularization of isolated disease of the left anterior descending coronary artery. Investigators in the Medicine, Angioplasty or Surgery Study (MASS) used a combined endpoint of cardiac death, MI, or refractory angina requiring repeat revascularization by surgery; at 3 years of follow-up, this combined endpoint occurred in 24% of PTCA patients, in 17% of medical patients, and in 3% of surgical patients ⁽¹⁶²⁾. Importantly, there was no difference in overall survival in the 3 groups. In the Lausanne trial of 134 patients with isolated left anterior descending artery disease treated by either PTCA (68 patients) or bypass with an internal mammary artery, survival was similar in the 2 groups, and 94% of PTCA patients and 95% of CABG patients were free of limiting symptoms ⁽¹⁶³⁾. However, patients in the PTCA group took more antianginal drugs than surgical patients, and at median follow-up of 2.5 years, 86% of CABG-treated versus 43% of PTCA-treated patients were free from late events (p < 0.01); this difference was primarily due to restenosis (32%) requiring subsequent CABG (16%) or PTCA (15%). It should be emphasized that neither of the 2 aforementioned trials included stenting, a technique which would be expected to reduce rates of early restenosis by as much as 50% in appropriately selected lesions ^(86,164,165).

In a similar manner, the 3-year follow-up of the Argentine randomized trial of PTCA versus CABG multivessel disease (ERACI study) ⁽¹⁶⁴⁾ demonstrated that in patients randomized to angioplasty or bypass surgery, the 1-, 3-, and 5-year follow-up results indicated that freedom from combined cardiac events was significantly greater for bypass surgery than for angioplasty group (77% vs. 47%; p < 0.001). However, there were no differences in overall and cardiac mortality or in the frequency of myocardial infarction between the two groups. Patients who had bypass surgery were more frequently free of angina (79% vs. 57%) and had fewer additional reinterventions (6.3% vs. 37%) than in patients who had angioplasty. This study indicated that freedom from combined cardiac events at 3-year follow-up was greater in bypass patients than those who had angioplasty and that the angioplasty group had a higher incidence of recurrence of angina and need for repeat procedures. Cumulative cost at 3 years was greater for surgery than for the angioplasty group.

In the ARTS trial, the first trial to compare stenting with surgery, there was no significant difference in mortality between PCI and surgical groups at one year. The main difference compared to previous PTCA and CABG trials was an approximate 50% reduction in the need for repeat revascularization in a group randomized to PCI with stent placement ⁽¹⁶⁶⁾.

Direct comparison of initial strategies of PCI or CABG in patients with multivessel coronary disease is possible only by randomized trials because of selection criteria of patients for PCI. There have been 5 large (>300 patients) randomized trials of PTCA vs. CABG and 2 smaller studies; characteristics of the studies are summarized in Table 10 ^{(9-12,164,167,168)}. These trials demonstrate that in appropriately selected patients with multivessel coronary disease, an initial strategy of standard PTCA yields similar overall outcomes (e.g., death, MI) compared to initial revascularization with coronary artery bypass. In BARI, the only trial with the largest patient enrollment to look at survival alone, 5-year survival was 86.3% for those assigned to PTCA vs. 89.3% for those assigned to CABG (p = 0.19), and 5-year survivals free from Q-wave MI were 78.7% and 80.4%, respectively. However, after 5 years of follow-up, 54% of those assigned to PTCA had undergone additional revascularization procedures compared to 8% of the patients assigned to CABG ⁽⁹⁾. Indications for PCI for various patient subsets are presented in Section V. Indications.

An important exception to the conclusion of the relative safety of PCI in multivessel disease is the subgroup of patients with treated diabetes mellitus. Among treated diabetic patients in BARI assigned to PTCA, 5-year survival was 65.5% compared to 80.6% for patients having CABG (p = 0.003); the improved outcome with CABG was due to reduced cardiac mortality (5.8% vs. 20.6%, p = 0.0003), which was confined to those receiving at least 1 internal mammary artery graft ⁽⁹⁾. Better survival of diabetic patients with multivessel disease treated initially with CABG has been observed in a large retrospective study from Emory ⁽¹⁶⁹⁾ and may be due to the apparent additive effects of diabetes mellitus and instrumentation of an artery on development of new stenotic lesions ⁽¹⁷⁰⁾. As compelling as these reports may be, it is of interest that treated diabetic patients enrolled in the BARI Registry did not show a similar advantage for CABG over PCI, suggesting that physician judgment in the selection of diabetic patients for PCI may be an important factor ^(38,48).

Moreover, direct comparison between outcomes of PCI and CABG among the diabetic population has not been made using platelet receptor antagonists with PCI. In this setting, PCI may be more competitive with CABG. The EPISTENT trial demonstrated significant reductions of major cardiac events at 30 days and at 6 months in the abciximab groups undergoing stenting compared to those with stenting and placebo ⁽¹³³⁾.

Randomized trials of PTCA and CABG provide additional information on symptom relief, quality of life, and costs of the 2 revascularization methods. Both revascularization techniques relieve angina. However, to achieve similar clinical outcomes, patients treated with PTCA are more likely to require further interventions than patients having surgery. Analysis of quality-of-life data from BARI suggests that functional status including activities of daily living improved less in patients assigned to PTCA than in those assigned to CABG (p < 0.05), although patients with initial PTCA returned to work 5 weeks sooner than did patients undergoing operation (p < 0.001) ⁽¹⁷¹⁾.

G. Comparison With Medicine

There has been a considerable effort made to evaluate the relative effectiveness of bypass surgery as compared to PCI for coronary artery revascularization. In contrast to this, very little effort has been directed toward comparing medical therapy with PCI for the management of stable and unstable angina. 3 Randomized trials are currently available comparing PCI with the medical management of angina ^(172-174). The ACME investigators randomized 212 patients with single-vessel disease, stable angina pectoris, and ischemia on treadmill testing to PTCA or medical therapy. This trial demonstrated superior control of symptoms and better exercise capacity in patients managed with PTCA as compared to medical therapy. Death and MI were infrequent and similar in both groups. The Veterans Administration ACME trial investigators long-term results in an additional 101 randomized patients with double-vessel disease not previously reported ⁽¹⁷⁵⁾ indicated that patients randomized to medical therapy or PTCA had similar improvement in exercise duration, freedom from angina, and improvement in quality of life at the time of 6-month follow-up. Thus, these patients with double-vessel angioplasty did not demonstrate superior control of their symptoms as compared to medical therapy as was experienced by the ACME patients with single-vessel disease. This small study suggests that PTCA is less effective in controlling symptoms in patients with double-vessel and stable angina as compared to single-vessel disease.

The RITA-2 investigators randomized 1018 stable patients with stable angina to PTCA or conservative (medical) therapy ⁽¹⁷³⁾. Patients who had inadequate control of their symptoms with optimal medical therapy were allowed to cross-over to myocardial revascularization. The combined endpoint of the trial was all cause mortality and nonfatal MI. The 504 PTCA and 514 medical patients were followed for a mean of 2.7 years. Death and definite MI occurred in 32 of the PTCA patients (6.3%) and in 17 of the medical patients (3.3%), p = 0.02. Of the 18 deaths (11 PTCA and 7 medical) only 8 were due to heart disease. Twenty-three percent of the medical patients required a revascularization procedure during follow-up. Angina improved in both groups, but there was a 16.5% absolute excess of grade 2 or worse angina in the medical group at 3 months following randomization (p < 0.001). The PTCA patients also had greater improvement in their exercise duration as compared to the medical patients (p < 0.001). During follow-up 40 patients randomized to PTCA required CABG surgery (7.9%) as compared to 30 of the medical patients (5.8%). Thus, RITA-2 demonstrated that PTCA results in better control of symptoms of ischemia and improves exercise capacity as compared to medical therapy, but is associated with a higher combined endpoint of death and periprocedural MI. It is important to remember that although the patients in this trial were asymptomatic or had only mild angina, 62% of them had multivessel CAD and 34% had significant disease in the proximal segment of the left anterior descending coronary artery ⁽¹⁷⁶⁾. Thus, most of these patients had severe anatomic CAD.

The Asymptomatic Cardiac Ischemia Pilot (ACIP) study provides additional information comparing medical therapy with PTCA or CABG revascularization in patients with documented CAD and asymptomatic ischemia by both stress testing and ambulatory ECG monitoring ⁽¹⁷⁶⁾. This trial randomized 558 patients suitable for revascularization by PTCA or CABG to 3 treatment strategies: angina-guided drug therapy (n = 183), angina plus ischemia-guided drug therapy (n = 183), and revascularization by PTCA or CABG surgery (n = 192). Of the 192 patients that were randomized to revascularization, 102 were selected for PTCA and 90 for CABG. At 2 years of follow-up, death or MI had occurred in 4.7% of the revascularization patients as compared to 8.8% of the ischemia-guided group and 12.1% of the angina-guided group (p < 0.01). Because a large portion of the patients underwent CABG surgery instead of PTCA in order to achieve complete revascularization, it is not appropriate to directly compare these results with RITA-2. Nonetheless, the ACIP study suggests that outcomes of revascularization with CABG surgery and PTCA are very favorable compared to medical therapy in patients with asymptomatic ischemia with or without mild angina. It should be emphasized that aggressive lipid-lowering therapy was not widely employed in the medical treatment arm of ACIP.

AVERT ⁽¹⁷⁴⁾ randomly assigned 341 patients with stable CAD, normal LV function, and Class I and/or II angina to PTCA or medical therapy with 80 mg daily atorvastatin (mean LDL = 77 mg/dL). At 18 months follow-up, 13% of the medically treated group had ischemic events as compared to 21% of the PTCA group (p = 0.048). Angina relief was greater in those treated with PTCA. Although not statistically different when adjusted for interim analysis, these data suggest that in low-risk patients with stable CAD, aggressive lipid-lowering therapy can be as effective as PTCA in reducing ischemic events.

Based on the limited data available from randomized trials comparing medical therapy with PTCA, it seems prudent to consider medical therapy for the initial management of most patients with Canadian Cardiovascular Society Classification Class I and II and reserve PTCA and CABG for those patients with more severe symptoms and ischemia. The symptomatic individual patient who wishes to remain physically active, regardless of age, will more often require PCI although one trial (RITA-2) ^(94,173) suggests that this option may be associated with an increased initial risk. The results of the ACIP trial indicate that higher-risk patients with asymptomatic ischemia and significant CAD who undergo complete revascularization with CABG or PTCA may have a better outcome as compared to those with medical management. This finding had not been previously demonstrated by trials comparing medical management with surgical revascularization ^(16,98) (Table 11). In contrast, the results of AVERT indicate revascularization provides no benefit when compared to aggressive lipid-lowering therapy in low-risk patients. Clinical Outcomes Utilization Revascularization and Aggressive Drug Evaluation (COURAGE) trial, a 3250 patient-based trial, will compare intensive medical therapy with revascularization over 5 to 7 years. It is anticipated that this trial will answer many questions, in addition to quality-of-life assessment and economic cost analysis ^(177-179) Patients with unstable angina and non-ST-segment elevation MI have been randomized to medical therapy or PCI in the FRISC II and TACTICS TIMI 18 trials. These trials utilizing stenting as the primary therapy have favored the invasive approach. They are discussed under Section V. B.