The Changing Pattern of Coronary Perforation During Percutaneous Coronary Intervention in the New Device Era

ABSTRACT: We report the incidence, management and clinical outcome of coronary perforations in 39 of 12,658 patients (0.3%) undergoing percutaneous coronary intervention (PCI). Coronary perforation occurred more frequently with debulking techniques than with non-debulking (percutaneous transluminal coronary angioplasty and stent) techniques (1% versus 0.2%; p 1–5 The incidence of coronary perforation is low, occurring in less than 0.2% of patients following balloon angioplasty (percutaneous transluminal coronary angioplasty), and seems to occur more frequently with the use of new coronary devices, such as high-speed rotational atherectomy, directional coronary atherectomy (DCA), laser and stents.^1–5 Furthermore, its magnitude and management may be further complicated with the use of glycoprotein (GP) IIb/IIIa platelet inhibitors, which are being used with increasing frequency during PCI.4 The purpose of the present study is to report the incidence, management and clinical outcome of coronary perforations occurring in patients undergoing PCI in the current device era. Methods Patient population. We identified 39 patients (0.3%) who developed coronary perforation during the PCI procedure out of 12,658 consecutive patients who underwent PCI at the Massachusetts General Hospital between January 1995 and March 2003. They comprise the patient population. Patient demographics, coronary risk factors, pre-procedural associated major comorbidities and clinical admission syndrome [including stable angina, unstable angina, post-myocardial infarction (MI) angina, evolving MI, cardiogenic shock and congestive heart failure] data were prospectively collected and entered in an interventional database. Congestive heart failure was assessed in accordance with the New York Heart Association classification. Renal disease was defined as a documented history of renal failure or a serum creatinine of > 1.5 mg/dl irrespective of the etiology. Hypertension was defined as blood pressure >= 160/95 mmHg or when subjects were chronically taking anti-hypertensive medication. Hypercholesterolemia was defined as serum cholesterol of greater than 200 mg% or chronic intake of lipid lowering drugs. Smoking was defined as current or former (> 1 month) use of nicotine. Family history was positive when a first degree relative (parent or sibling) had a documented history of coronary artery disease. Vascular disease was defined as the presence of peripheral or cerebrovascular disease. Lesion morphology was classified according to the American Heart Association/American College of Cardiology (AHA/ACC) Classification Task Force, with the exception that type B lesions were further stratified into B1 and B2 lesions according to Ellis et al.^6,7 All patients underwent PCI according to practice guidelines at the time of the procedure. Adjunctive therapies included aspirin, ticlopidine or clopidogrel, and GP IIb/IIIa antagonists. Conventional steerable guidewire systems were used, and the operator selected the interventional devices at the time of the intervention. Patients were examined angiographically before and immediately after coronary intervention. Similar single view projections were used at each angiographic examination. Percent coronary stenosis, reference and minimal luminal diameters were determined using quantitative coronary analysis after intracoronary administration of 100 µg nitroglycerin using a computer-assisted, automated edge detection algorithm (Computer Measurements System, MEDIS, Nuenen, The Netherlands).^8,9 Absolute reference and minimal luminal diameters were determined (in millimeters) using the guiding catheter filled with contrast for calibration. For each lesion, the single view showing the most severe degree of stenosis was used for analysis. A lesion treatment was considered to be successful when there was >= 20% gain in luminal diameter and 3 type I = crater extending outside the lumen only in the absence of linear staining angiographically suggestive of a dissection; type II = pericardial or myocardial blush without an exit hole >= 1 mm; type III = frank streaming of contrast through an exit hole >= 1 mm; and type III cavity spilling = perforation into an anatomic cavity chamber such as the coronary sinus, the right ventricle, etc. All patients remained in the hospital for a minimum of 24 hours after the procedure. In-hospital major adverse cardiac events (MACE) included death (cardiac and non-cardiac) or Q-wave MI occurring during the hospital stay irrespective of the time elapsed between the procedure and the event, or emergency coronary artery bypass surgery performed within the first 24 hours after the procedure. Patient charts were reviewed daily for adverse events until hospital discharge or death. Statistical analysis. Continuous and categorical variables were compared by student’s t-test and Chi-square analysis, respectively. Data are presented as means ± standard deviation and p-values Patient population. From January 1995 to March 2003, coronary perforation occurred in 39 out of 12,658 consecutive patients (0.3%) who underwent PCI at Massachusetts General Hospital. Coronary perforation occurred in 16 of the 2,616 patients who underwent PTCA (0.61%), in 13 of the 8,281 patients who underwent stent procedures (0.15%) and 10 of the 1,005 patients who underwent debulking procedures (rotational and directional coronary atherectomy) (1%). Thus, coronary perforation occurred more frequently with debulking techniques than with non-debulking (PTCA and stent) techniques (1% versus 0.26%; p Debulking techniques and severity of coronary perforation. Ten patients suffered a coronary perforation while undergoing a debulking technique; three patients (30%) had a type I perforation, three patients (30%) had a type II perforation and 4 patients (40%) had a type III perforation. The likelihood of developing a type III perforation was not significantly greater in the debulking group versus the non-debulking group. GP IIb/IIIa inhibitors in the setting of coronary perforation during PCI. Of the 12,658 patients undergoing PCI, a total of 6,067 (48%) received GP IIb/IIIa inhibitors. Coronary perforation occurred in 16 of these patients, compared with 23 of the 6,067 patients who did not receive GP IIb/IIIa inhibitors (0.26% versus 0.3%; p = NS). Of the 16 patients who had perforation while on GP IIb/IIIa inhibitors, type I perforation occurred in 4 patients, type II occurred in 3 patients and type III occurred in 9 patients. Finally, cardiac tamponade occurred in 4 of 16 patients receiving GP IIb/IIIa inhibitors, compared with 3 of 23 patients who did not receive GP IIb/IIIa inhibitors (25% versus 13%; p = 0.91). Treatment strategy. The treatments received by patients with coronary perforation are shown in Table 3. Immediate balloon occlusion of the vessel perforation site was performed in 28 patients (in 2 patients with type I perforation and in all patients with type II and type III perforations). A perfusion balloon was used in 8 patients (20.5%). A covered stent was used in 2 patients with type III perforations who continued to bleed after prolonged balloon inflation and reversal of the anticoagulation. Reversal of heparin anticoagulation with protamine sulfate was performed in 11 patients (28.2%). Intravenous administration of platelets was required in 1 of 16 patients who received GP IIb/IIIa platelet antagonists. In-hospital outcomes. In-hospital outcomes are shown in Table 4. Despite coronary perforation, a successful interventional outcome was obtained in 28 patients (71.7%). There was 1 death (2.5%), an 84-year old man who developed electromechanical dissociation and hypotension immediately after directional coronary atherectomy and stenting of a left anterior descending/diagonal bifurcation lesion, and the autopsy demonstrated coronary perforation and hemopericardium. Two patients (5%) required emergency surgery and no patient suffered a Q-wave MI. A non-Q wave MI occurred in 7 patients (17.9%). Pericardial effusion developed in 14 of 39 patients (36%): one with type I perforation, three with type II perforation and 10 with type III perforation. Eight of these patients (57%) had guidewire perforation. Cardiac tamponade developed in 7 of 16 patients with type III perforation, but in no other perforation group. Guidewire perforation was the cause of cardiac tamponade in 3 cases. Finally, one patient developed severe hemoptysis following a type III perforation of a saphenous vein bypass graft into a right-sided bronchus. Discussion The present study demonstrated that the incidence of coronary perforation during PCI in the new device era characterized by high usage of secondary devices and GP IIb/IIIa platelet antagonist remains low. The 0.3% incidence of coronary perforation in the present study is similar to the 0.29%, 0.58% and 0.48% reported by the Washington Hospital Medical Center, the Ohio Heart Health Center and the Cleveland Clinic, respectively.^3–5 In agreement with those studies, our study demonstrated that the incidence of coronary perforation was greater with the use of debulking procedure, such as directional coronary atherectomy and rotational atherectomy, than with the use of non-debulking procedures such as PTCA and coronary stenting. However, our findings disagree with previous reports regarding the likelihood of severe coronary perforation with the use of debulking techniques,^3,4 as 40% of our patients with debulking techniques versus 55% with non-debulking techniques had type III perforation (p = NS). Other risk factors for coronary perforation are related to complex coronary anatomy (calcified lesion, chronic total occlusion, tortuosity of the vessel and ostial lesion). Similar to previous reports, most of our patients had complex anatomy (American College of Cardiology/American Heart Association lesion class B and C), making the intervention more difficult. Our study concluded that distal migration of the guidewire in the presence of GP IIb/IIIa inhibition is an important factor for coronary perforation. Therefore, meticulous care of the guidewire should be taken when PCI is performed in these complex lesions, as 51% of the coronary perforations were guidewire-related in the present study. Although coronary perforation can occur while attempting to cross the lesion with the guidewire, the distal wire by itself or a fracture guidewire can often cause the perforation. The importance of this finding cannot be overemphasized, as 9 (45%) and 7 (35%) of the 20 guidewire perforations were type II and III, respectively. Moreover, pericardiocentesis for relief of cardiac tamponade was required in 3 of the 7 patients (43%) with guidewire type III perforation. Perforation due to excimer laser coronary angioplasty has been previously described;¹⁰ however, none of our PCI patients underwent laser procedures during this period. It is also important to recognize that perforation is not always immediately evident, as tamponade can present later. This diagnosis should remain high on the list of the differential diagnosis of post-PCI hypotension. Another form of cardiac perforation not included in the present study occurs as a consequence of the use of a temporary pacing wire, and interventional cardiologists should be aware of this cause of pericardial effusion and tamponade in patients undergoing PCI.¹¹ The incidence of perforation in patients who received GP IIb/IIIa inhibitors in the present study (0.26%), is similar to that reported with the use of abciximab in the EPIC, EPILOG, CAPTURE and EPISTENT trials (Figure 2).^12–15 Overall, coronary perforation occurred in 7 of 3,054 patients (0.2%) in the placebo arms and in 22 of 5,436 patients (0.4%) treated with abciximab (p = 0.24). Our study further supports the concept that major adverse clinical outcomes are related to the angiographic classification of perforation, occurring more frequently in patients who experienced type III perforation.³ In the present study, we found a trend for an increased incidence of cardiac tamponade in patients receiving GP IIb/IIIa antagonists. However, as reported by Dippel et al., their use was not associated with a deleterious impact on mortality or emergency surgery rates.⁴ Treatment of coronary perforation in the current PCI era, which is characterized by high utilization of new devices and GP IIb/IIIa platelet antagonists, requires early detection, angiographic classification, immediate balloon occlusion of coronary vessel extravasation and relief of hemodynamic compromise (Figure 3). Keeping in mind that pericardial effusion and tamponade are more likely to occur in type II and III perforations, immediate occlusion of the perforated vessel should be accomplished by prompt and prolonged balloon catheter inflation at the perforation site. Perfusion balloon catheters can be used for prolonged balloon occlusion, maintaining myocardial perfusion and therefore avoiding myocardial ischemia during balloon occlusion. Immediate attention should be directed to reversal of heparin anticoagulation. This is accomplished with the intravenous administration of protamine sulfate to achieve a partial thromboplastin time of less than 60 seconds or an ACT of less than 150 seconds. Eleven of our patients received protamine as part of their treatment and none of them had vessel thrombosis complications. As previously described, the use of protamine in patients with coronary perforation post-PCI seems to be safe.¹⁶ Platelet transfusion is useful in patients treated with abciximab, but not with tirofiban or eptifibatide. A major development in the treatment of coronary perforation in the current PCI era is the use of covered stents. Polytetrafluoroethylene (PTFE) covered stents can successfully seal the site of coronary perforation, and therefore decrease mortality and need for emergency surgery in these patients.¹⁷ Our study includes patients with coronary perforation and spans 8 years when covered stents were not available. This finding is responsible for the low frequency of Jomed stent utilization in the present study. Other forms of sealing perforated vessels described prior to the introduction of the covered stent include the use of autologous vein graft,¹⁸ makeshift stent sandwich19 and microcoil embolization.²⁰ None of these methods were employed in our patient population. Conclusion. The incidence of coronary perforation remains low in the current device era; it occurs more frequently with debulking devices and often as a consequence of guidewire migration and injury in the presence of GP IIb/IIIa inhibition. Nevertheless, its outcome is not affected by the use of GP IIb/IIIa antagonists. Treatment of coronary perforation requires early detection, angiographic classification, immediate occlusion of coronary vessel extravasation and relief of hemodynamic compromise, reversal of heparin anticoagulation, platelet transfusion in those patients treated with abciximab, and the use of cover stents.