From the a Duke Clinical Research Institute, Durham, NC, Singapore, Singapore,

Transcription

1 Comparison of percutaneous coronary intervention for previously treated versus de novo culprit lesions in acute myocardial infarction patients: insights from the National Cardiovascular Data Registry Chee Tang Chin, MBChB, MRCP(UK), a,b John C. Messenger, MD, c David Dai, PhD, a Lisa A. McCoy, MS, a Michael A. Kutcher, MD, d H. Vernon Anderson, MD, e Matthew T. Roe, MD, MHS, a and Tracy Y. Wang, MD, MHS, MSc a Durham, and Winston-Salem, NC; Aurora, CO; Singapore, Singapore; and Houston, TX Background Little is known about percutaneous coronary intervention (PCI) outcomes among patients presenting with an acute myocardial infarction (MI) with a history of prior PCI. Outcomes may differ depending on whether PCI is performed on a previously treated or de novo culprit lesion. Methods We examined ST-segment elevation myocardial infarction (STEMI) and non-stemi patients who underwent PCI in the CathPCI Registry from 2009 to We used multivariable logistic regression to compare adjusted in-hospital mortality between groups. Results Among 675,587 MI patients, 147,841 (22%) had a history of prior PCI; these patients were older and more frequently had co-morbid conditions yet had lower adjusted mortality compared with patients undergoing their first intervention (OR = 0.73, 95% CI = ). Among patients with prior PCI, 50,744 (34%) received intervention to a culprit lesion in a previously treated segment. Compared with patients with de novo culprit lesions, those with previously treated culprits were more likely to present with STEMI, but had lower mortality risk (OR = 0.88, 95% CI = ) regardless of STEMI or non- STEMI presentation. Among previously treated patients, in-hospital mortality was not significantly different between those with prior drug-eluting versus bare metal stent-treated culprit lesions (OR = 0.95, 95% CI = ). Conclusion Despite greater co-morbidity burden, MI patients with prior PCI had lower mortality compared with patients undergoing their first intervention. Among patients with prior PCI, patients undergoing PCI to a previously treated culprit lesion were associated with lower mortality than those being intervened for a de novo culprit. A better understanding of these differences will help improve procedural strategies and outcomes of patients undergoing PCI of a previously treated lesion. (Am Heart J 2014;167: e1) From the a Duke Clinical Research Institute, Durham, NC, b National Heart Centre Singapore, Singapore, c University of Colorado School of Medicine, Aurora, CO, d Wake Forest University School of Medicine, Winston-Salem, NC, and e University of Texas Health Science Center, Houston, TX. Mauricio G. Cohen, MD, served as guest editor for this article. Submitted October 1, 2013; accepted December 5, Reprint requests: Chee Tang Chin, MBChB, MRCP(UK); 17 Third Hospital Avenue; Singapore chin.chee.tang@nhcs.com.sg /$ - see front matter 2014, Mosby, Inc. All rights reserved. Percutaneous coronary intervention (PCI) with stenting is an established revascularization strategy for patients with both stable and acute coronary syndromes. 1 Previously stented coronary lesions are at risk of instent restenosis or stent thrombosis, with pathophysiologic underpinnings that are distinct from each other and from de novo coronary lesions. 2,3 Yet little is known about PCI outcomes for patients with prior PCI who present with acute myocardial infarction (MI). We hypothesized that treatment strategies and outcomes may differ depending on whether PCI is performed on a previously treated or de novo lesion, and whether the lesion was previously treated with a drug-eluting stent (DES) or bare metal stent (BMS). Any such differences have not been previously characterized. Focusing on patients presenting with either ST-segment elevation myocardial infarction (STEMI) or non-stemi (NSTEMI) in the National Cardiovascular Data Registry CathPCI Registry, we compared outcomes between: (1) PCI of patients with and without prior PCI; (2) PCI treating a previously treated versus de novo culprit lesion among patients with prior PCI; and (3) PCI treating a

2 394 Chin et al American Heart Journal March 2014 Figure 1 Study population. Flowchart showing, after study inclusions and exclusions, the final study population of 675,587 MI patients undergoing PCI at 1,391 sites. culprit lesion previously stented with DES versus BMS among MI patients with previously treated lesions. Methods Data source The CathPCI Registry is the largest registry of patients undergoing PCI in the United States. Co-sponsored by the American College of Cardiology and the Society for Cardiovascular Angiography and Interventions, the registry has been well described previously ( DefaultCathPCI.aspx). 4,5 Briefly, participating centers submit complete information from consecutive PCI cases performed. As patient information was collected anonymously without unique patient identifiers, individual informed consent was not required. Participation in the CathPCI Registry was subject to the approval of the institutional review board of each hospital. Patient characteristics, presentation features, angiographic and procedural details, and in-hospital outcomes were collected using standardized data elements and definitions, as described previously. 5 Measures such as rigorous and uniform data abstraction training, point-of-entry data quality threshold verification, site feedback reports, independent auditing, and data validation ensure that data quality is maintained. Study population Our starting population included 770,529 STEMI and NSTEMI patients treated with PCI at 1,392 United States hospitals in the CathPCI Registry from July 1, 2009, through December 31, We excluded patients with previous coronary artery bypass graft surgery (n = 94,942). For MI patients with multiple catheterization laboratory visits during the MI hospitalization, we analyzed only the first PCI after admission as this initial PCI most likely involved the culprit lesion. This yielded a final study population of 675,587 patients with native coronary artery disease undergoing PCI for acute MI. Data definitions We defined MI patients as all patients who presented with STEMI or NSTEMI. The CathPCI Registry data collection form captures the culprit lesion for the MI. If no culprit was designated, then the first lesion treated was considered the culprit lesion. The data collection form also captured whether the intervened lesion was stented during a previous PCI procedure, and whether DES or BMS was used in the prior stenting procedure. We examined outcomes of in-hospital mortality and PCI procedural success. Procedural success was defined as successful dilation of all lesions attempted during the same setting with post-procedure TIMI 3 flow, post-procedure stenosis 50%, and a decrease between pre- and post-procedure percent stenosis 20%. All data element definitions used by the CathPCI Registry are available at webncdr/elements.aspx. This work was supported by funding from the American College of Cardiology National Cardiovascular Data Registry. Statistical methods We first divided the study population into PCI patients with and without a history of prior PCI. Among prior PCI patients, we then divided patients into those undergoing PCI of a previously

3 American Heart Journal Volume 167, Number 3 Chin et al 395 treated or a de novo culprit lesion. Finally, among patients undergoing PCI of a previously treated lesion with known prior stent type, we classified patients based on whether the culprit lesion had been previously stented with a DES or a BMS. Patient characteristics, angiographic findings, management strategies, complications, and outcomes were compared using Pearson chi-square tests for categorical variables, and chisquare rank-based group means score statistics for continuous and ordinal variables. Our primary outcome of interest was inhospital mortality. A secondary outcome of interest among prior PCI patients was PCI procedural success comparing previously treated with de novo culprit lesions, and comparing prior DES-treated with prior BMS-treated lesions. To account for within-hospital clustering and differences in patient characteristics, the multivariable logistic generalized estimating equations method was used to generate a multivariable model comparing risk-adjusted outcomes between groups. 6 Variables used in the model were adapted from the validated CathPCI Registry in-hospital mortality model (c-index 0.92) and are listed in the online Appendix. 7 We then tested the interaction of STEMI or NSTEMI presentation on outcomes among the different groups. All comparisons were 2-tailed, and P b.05 was considered statistically significant. All analyses were performed by the CathPCI Registry data analysis center at the Duke Clinical Research Institute using SAS version 9.2 (SAS Institute, Cary, NC). Results Between July 1, 2009, and December 31, 2012, 675,587 patients underwent PCI for an acute MI at 1,391 hospitals in the United States. As shown in Figure 1, 147,841 (21.9%) patients had a history of prior PCI, while this was the first coronary intervention for 527,746 (78.1%) patients. Among patients with a prior PCI, the culprit lesion was identified in a previously-stented coronary segment in 50,744 (34.3%) patients and in a de novo segment in 97,097 (65.7%) patients. No prior PCI versus prior PCI Patients with prior PCI were older and more likely to have prior cardiovascular disease (prior MI, heart failure, cerebrovascular disease, and peripheral artery disease), as well as cardiovascular risk factors (diabetes, dyslipidemia, and hypertension) than patients undergoing their first intervention (Table I). Prior PCI patients were less likely to present with STEMI (42% vs 52%, P b.001) but more likely to have heart failure on presentation compared with patients without prior PCI. Prior PCI patients were less likely to have occlusive flow or culprit lesions in the proximal segments of the major epicardial vessels; 15% of these patients underwent balloon-only PCI without new stenting. Among stented patients, those who had prior PCI were more likely to receive a DES than patients without prior PCI. Glycoprotein IIb/IIIa inhibitors were used less frequently among prior PCI patients. Table I. Comparison of patients with prior PCI versus no prior PCI No prior PCI (n = 527,746) Prior PCI (n = 147,841) P Baseline characteristics Age (years) 61 (52,71) 63 (54,73) b.001 Female b.001 Prior myocardial b.001 infarction Prior heart failure b.001 Cerebrovascular disease b.001 Peripheral artery disease b.001 Diabetes b.001 Hypertension b.001 Dyslipidemia b.001 Current smoking b.001 Presenting features STEMI (vs NSTEMI) b.001 Heart failure b.001 Cardiogenic shock b.001 Creatinine clearance 69 (50,89) 64 (45,85) b.001 (ml/min), Dialysis b.001 Angiographic characteristics Culprit location b.001 Left main Proximal LAD Prox RCA/Prox Cx/Mid LAD Others Pre-procedure b.001 TIMI flow 0/1 Thrombus noted b.001 Lesion length (mm) 18 (13,24) 18 (12,24) b.001 Bifurcation lesion b.001 Treatment characteristics Balloon only (no stent) b.001 % DES used (among b.001 stented) Thrombectomy b.001 Multivessel PCI b.001 IABP used b.001 Periprocedural pharmacology Unfractionated heparin b.001 Low molecular weight b.001 heparin GpIIb/IIIa inhibitor b.001 Bivalirudin b.001 Cx, Circumflex artery; GpIIb/IIIa, glycoprotein IIb/IIIa inhibitor; IABP, intra-aortic balloon pump; LAD, left anterior descending; RCA, right coronary artery. Data presented as percentages except for: Where presented as median (25th percentile, 75th percentile). For non-dialysis patients. Despite greater comorbidity burden, in-hospital mortality risk was slightly lower in prior PCI patients compared with patients undergoing their first PCI (3.2% vs 3.7%), which persisted after multivariable adjustment (odds ratio [OR] 0.73, 95% confidence interval [CI] ). This association held for both STEMI and NSTEMI patients (P interaction =.24, Figure 2A).

4 396 Chin et al American Heart Journal March 2014 Figure 2 In-hospital mortality. After multivariable risk adjustment, in-hospital mortality (overall, and stratified by STEMI and NSTEMI) for patients with: prior PCI versus without prior PCI (A); previously treated versus de novo culprit lesions (B); and prior DES versus prior BMS culprit lesions (C). Previously-treated versus de novo culprit lesions Among MI patients with prior PCI, those who underwent PCI of a previously treated culprit lesion were younger, more likely to be smokers, and to have had a prior MI compared with patients undergoing PCI of a de novo culprit lesion (Table II). Patients in the previously treated group were significantly more likely to present with STEMI (51% vs 38%, P b.001) compared with de novo culprit patients. At angiography, patients undergoing PCI of a previously treated lesion were more likely have the culprit lesion located in the proximal coronary artery segments, and thrombus was more frequently observed in the culprit lesion. Compared with de novo group patients, previously treated patients were more likely to be treated with balloon angioplasty alone (28% vs 8%, P b.001); however, if stented, they were more likely to receive DES. Previously-treated patients were less likely to undergo multivessel PCI in the acute setting, have greater periprocedural use of glycoprotein IIb/IIIa inhibitors, and lower use of bivalirudin. Patients who underwent PCI of a culprit lesion in a previously treated segment had lower adjusted in-hospital mortality risk (OR 0.88, 95% CI ) compared with de novo culprit patients. This association was consistent among both STEMI and NSTEMI patients (P interaction =.42, Figure 2B). Percutaneous coronary intervention of a previously treated culprit lesion was associated with a modestly higher odds of procedural success compared with a de novo lesion (adjusted OR 1.05, 95% CI ). This was observed primarily among STEMI (adjusted OR 1.15, 95% CI ), but not NSTEMI patients (P interaction b.001, Figure 3A). Prior DES-treated lesion versus prior BMS-treated lesion Among acute MI patients who underwent PCI of a previously-stented culprit lesion, the lesion was previously stented with a DES in 24,085 patients (47%). Patients previously stented with DES were more likely to have diabetes, but less likely to have had prior MI compared with prior BMS patients (Table III). Patients previously stented with DES were more likely to present with STEMI compared with those previously stented with BMS. Prior DES patients were more likely to have observed thrombus and occlusive pre-procedure flow compared with prior BMS patients. Prior DES patients were more likely to undergo balloon angioplasty only without new stent implantation (30% vs 25%, P b.001), and if stented were significantly more likely to receive another DES (83% vs 69%, P b.001) compared with patients previously treated with BMS. In-hospital mortality was not significantly different between the 2 groups (adjusted OR 0.95, 95% CI ). Procedural success rates were also not significantly different between PCI of a lesion previously stented with

5 American Heart Journal Volume 167, Number 3 Chin et al 397 Table II. Comparison of previously treated versus de novo culprit lesion among patients with prior PCI Previously-treated (n = 50,744) de novo (n = 97,097) P Baseline characteristics Age (years) 62 (53,71) 64 (55,74) b.001 Female Prior myocardial b.001 infarction Prior heart failure Cerebrovascular disease b.001 Peripheral artery b.001 disease Diabetes b.001 Hypertension Dyslipidemia b.001 Current smoking b.001 Presenting features STEMI (vs NSTEMI) b.001 Heart failure b.001 Cardiogenic shock b.001 Creatinine clearance 65 (46,85) 64 (44,84) b.001 (ml/min), Dialysis b.001 Angiographic characteristics Culprit location b.001 Left main Proximal LAD Prox RCA/Prox Cx/Mid LAD Others Pre-procedure TIMI b.001 flow 0/1 Thrombus noted b.001 Lesion length (mm) 18 (12,25) 16 (12,23) b.001 Bifurcation lesion b.001 Treatment characteristics Balloon only b.001 (vs stent) % DES used b.001 (among stented) Thrombectomy b.001 Multivessel PCI b.001 IABP used b.001 Periprocedural pharmacology Unfractionated b.001 heparin Low molecular b.001 weight heparin GpIIb/IIIa b.001 Bivalirudin b.001 All abbreviations can be found in Table I. Data presented as percentages except for: Where presented as median (25th percentile, 75th percentile). For non-dialysis patients. a DES versus a lesion previously stented with a BMS (adjusted OR 1.01, 95% CI ). The association of prior DES versus BMS with outcomes was not significantly different between STEMI and NSTEMI patients (P interaction =0.65 for mortality [Figure 2C] and 0.34 for procedural success [Figure 3B]). Discussion Several insights emerge from this study. First, STEMI and NSTEMI patients with prior PCI, despite a higher risk profile, had lower in-hospital mortality than patients without prior PCI. Second, PCI of culprit lesions in a previously treated segment was associated with greater odds of procedural success and lower mortality risk than PCI of de novo culprit lesions. Finally, among MI patients with a previously-stented culprit lesion, the odds of procedural success and mortality did not significantly differ between patients with prior DES or prior BMS. In our study, approximately 1 in 5 MI patients undergoing PCI of native coronary artery disease had a history of prior PCI. This is consistent with other contemporary registry data of MI patients such as the ACTION Registry-GWTG. 8 Compared with patients undergoing their first PCI, those in our study with a history of prior PCI were older, more frequently diabetic, and were more likely to have had prior cardiovascular disease (such as MI, stroke, or peripheral arterial disease). Nevertheless, prior PCI patients were also more likely to have lower unadjusted in-hospital mortality rates which persisted after multivariable adjustment; we offer a few explanations for this unexpected finding. In part, this observation may be attributed to ischemic preconditioning as patients with prior PCI would more likely have been previously exposed to periods of myocardial ischemia than MI patients presenting for their first coronary intervention. Reports from animal studies as well as from patients undergoing PCI suggest that ischemic preconditioning may reduce infarct size and, therefore, may be associated with better outcomes. 9,10 Furthermore, prior PCI patients may have a better developed collateral circulation than patients with no prior PCI. The presence of collaterals has been associated with less myocardial ischemia, and reduced infarct size. 11 Furthermore, in a non-randomized observational setting, we cannot account for treatment bias where the treating physician may have pursued PCI only among selected high-risk prior PCI patients. Alternatively, the highest risk prior PCI patients may not have survived to repeat PCI, or may have had disease that was no longer amenable to PCI. Additionally, interventions on higher risk prior PCI patients may have been performed by more experienced proceduralists, with better resulting outcomes than a general MI population. Fineschi et al reported that among 85 DES patients who subsequently had clinically-driven repeat PCI for a new acute coronary syndrome event, 71 (78%) had culprit lesions identified in the previously-stented segment. 12 Our study expands on the existing literature by examining a large cohort of MI patients treated with PCI after a previous coronary intervention. In contrast to the

6 398 Chin et al American Heart Journal March 2014 Figure 3 Procedural success. After multivariable risk adjustment, procedural success (overall, and stratified by STEMI and NSTEMI) for patients with: previously treated versus de novo culprit lesions (A); and prior DES versus prior BMS culprit lesions (B). Fineschi study, we found that the majority of repeat PCI procedures in these MI patients were performed on de novo culprit lesions separate from their previously treated segments. Patients with a culprit lesion in a previously treated segment were more likely to have presented with STEMI than those with a de novo culprit lesion. Reassuringly, in our large study population, patients with a previously treated culprit lesion had a higher likelihood of procedural success and a lower risk of inhospital mortality than patients with de novo culprits in both STEMI and NSTEMI patients. However, more complex patients may have been referred for surgery or treated conservatively rather than with repeat PCI. We also sought to identify any differences in clinical presentation and PCI outcomes between a culprit lesion that had previously been treated with a BMS as opposed to a DES. Patients with a previously DES-treated segment were more likely to present with a STEMI, and thrombus was more commonly visualized, compared with prior BMS culprit patients. Despite this, procedural success and risk-adjusted mortality was similar between prior DES and prior BMS culprit groups. Another important observation from our study is the choice of treatment for MI patients with prior PCI. Among these patients, stents were used (as expected) in 89% of patients with de novo coronary lesions. In contrast, stenting was performed in only 67% of patients with prior DES and 72% of patients with a prior BMS. In certain situations, balloon angioplasty alone may be adequate to treat patients with previously-stented culprit lesions. Although one might expect higher rates of DES use among patients with prior BMS-treated culprits, the difference in DES use between prior BMS- and DEStreated groups was small. Of note, periprocedural glycoprotein IIb/IIIa inhibitor was used more frequently among MI patients with previously-stented culprit lesions with and without visible thrombus. Study limitations Some limitations should be considered in the interpretation of these results. First, the CathPCI Registry offers the advantage of a cross-sectional perspective of patients undergoing PCI at more than 1000 United States hospitals. Nevertheless, although the registry examines nationwide PCI outcomes among MI patients who have had prior coronary intervention, it is unable to follow patients longitudinally after stenting or assess patients who were treated medically or with coronary artery bypass graft surgery. Second, angiographic characteristics in the CathPCI Registry were collected by individual hospitals using standardized data definitions, but were not independently adjudicated by an angiographic core laboratory, as this is beyond the scope of this national registry. As a result, we could not assess the true incidence of Academic Research Consortium-defined stent thrombosis among MI patients with previous stenting, nor provide detailed lesion characteristics (eg, location within the body of the previous stent versus involving a stent edge). Third, details of the initial stent procedure beyond segment location and stent type are not captured in this registry. Conceivably, initial lesion characteristics, as well as factors related to initial stent implantation (such as severity of coronary calcification and degree of stent apposition) may influence downstream outcomes. Fourth, while the observational nature of this study permits real-world assessment of patient outcomes, the association with outcomes does not prove causality. Furthermore, the impact of operator skill or

7 American Heart Journal Volume 167, Number 3 Chin et al 399 Table III. Comparison of prior BMS versus prior DES among patients with previously treated lesion Prior DES (N = 24,085) Prior BMS (N = 10,120) P Baseline characteristics Age (years) 61 (53,71) 62 (53,72) b.001 Female Prior myocardial infarction b.001 Prior heart failure Cerebrovascular disease Peripheral artery disease Diabetes b.001 Hypertension b.001 Dyslipidemia b.001 Current smoking b.001 Presenting features STEMI (vs NSTEMI) b.001 Heart failure b.001 Cardiogenic shock Creatinine clearance (ml/min), (45.9, 84.9) (45.2, 86.4) Dialysis Angiographic characteristics Culprit location.002 Left main Proximal LAD Prox RCA/Prox Cx/Mid LAD Others Pre-procedure TIMI b.001 flow 0/1 Thrombus noted b.001 Lesion length (mm) 18 (12, 24) 20 (15, 26) b.001 Bifurcation lesion b.001 Treatment characteristics Balloon only (no stent) b.001 % DES used b.001 (among stented) Thrombectomy b.001 Multivessel PCI IABP used Periprocedural pharmacology Unfractionated heparin Low molecular weight heparin GP IIb/IIIa b.001 Bivalirudin Data presented as percentages except for: All abbreviations can be found in Table I. Where presented as median (25th percentile, 75th percentile). For non-dialysis patients. experience on outcomes cannot be assessed. Finally, only inhospital outcomes are reported in the CathPCI Registry and it is unknown if longer-term outcomes would differ between the groups. history of prior PCI. Despite greater co-morbidity burden, these MI patients with prior PCI had lower risk-adjusted in-hospital mortality. Among prior PCI patients, approximately one-third of patients had a culprit lesion that involved a previously treated coronary segment. Patients with culprit lesions in a previously treated segment were more likely to present with STEMI, but had lower inhospital mortality and greater procedural success rates compared with patients who had de novo culprit coronary lesions. There was no difference in mortality between patients with previous DES-treated culprit lesions and those who had previous BMS-treated culprits. Future studies to better understand these differences and mechanistic underpinnings will further improve interventional strategies and outcomes of patients undergoing intervention of a previously treated coronary lesion. Funding Sources The CathPCI Registry is an initiative of the American College of Cardiology Foundation and The Society for Cardiovascular Angiography and Interventions. Acknowledgements The authors would like to thank Erin Hanley, MS for her contributions to this manuscript. Ms. Hanley did not receive compensation for her assistance, apart from her employment at the institution where the study was conducted. Disclosures Conflict of Interest Disclosures: CT Chin: Dr Chin has no relevant conflicts of interest to report. JC Messenger: Dr Messenger has no relevant conflicts of interest to report. D Dai: Dr Dai has no relevant conflicts of interest to report. LA McCoy: Ms McCoy has no relevant conflicts of interest to report. MA Kutcher: Dr Kutcher has no relevant conflicts of interest to report. HV Anderson: Dr Anderson has no relevant conflicts of interest to report. MT Roe: Research grants: AstraZeneca, Bristol Myers Squibb, Eli Lilly and Company, Glaxo SmithKline, KAI Pharmaceuticals, Novartis, Sanofi-Aventis, Schering- Plough Corporation, and Orexigen. All conflicts of interest are listed at TY Wang: Research grants to the Duke Clinical Research Institute from Bristol- Myers Squibb/Sanofi Partnership; Schering Plough/ Merck, The Medicines Co, Heartscape, Canyon Pharmaceuticals, and Eli Lilly/Daiichi Sankyo Alliance, as well as consulting or honoraria for Medco and Astra Zeneca. All conflicts of interest are listed at Conclusions In this large contemporary cohort of patients undergoing PCI for STEMI or NSTEMI, one in five patients had a References 1. Kushner FG, Hand M, Smith Jr SC, et al focused updates: ACC/AHA guidelines for the management of patients with

8 400 Chin et al American Heart Journal March 2014 ST-elevation myocardial infarction (updating the 2004 guideline and 2007 focused update) and ACC/AHA/SCAI guidelines on percutaneous coronary intervention (updating the 2005 guideline and 2007 focused update) a report of the American College of Cardiology Foundation/American Heart Association task force on practice guidelines. J Am Coll Cardiol 2009;54: Holmes Jr DR, Kereiakes DJ, Garg S, et al. Stent thrombosis. J Am Coll Cardiol 2010;56: Dangas GD, Claessen BE, Caixeta A, et al. In-stent restenosis in the drug-eluting stent era. J Am Coll Cardiol 2010;56: Weintraub WS, McKay CR, Riner RN, et al. The American College of Cardiology National Database: progress and challenges. American College of Cardiology Database Committee. J Am Coll Cardiol 1997; 29: Brindis RG, Fitzgerald S, Anderson HV, et al. The American College of Cardiology-National Cardiovascular Data Registry (ACC-NCDR): building a national clinical data repository. J Am Coll Cardiol 2001; 37: Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics 1986;42: Peterson ED, Dai D, DeLong ER, et al. Contemporary mortality risk prediction for percutaneous coronary intervention: results from 588,398 procedures in the National Cardiovascular Data Registry. J Am Coll Cardiol 2010;55: Chin CT, Chen AY, Wang TY, et al. Risk adjustment for in-hospital mortality of contemporary patients with acute myocardial infarction: the acute coronary treatment and intervention outcomes network (ACTION) registry-get with the guidelines (GWTG) acute myocardial infarction mortality model and risk score. Am Heart J 2011;161: 113e2-22e2. 9. Murry CE, Jennings RB, Reimer KA. Preconditioning with ischemia: a delay of lethal cell injury in ischemic myocardium. Circulation 1986; 74: Heusch G. Cardioprotection: chances and challenges of its translation to the clinic. Lancet 2013;381: Charney R, Cohen M. The role of the coronary collateral circulation in limiting myocardial ischemia and infarct size. Am Heart J 1993;126: Fineschi M, Zacà V, Gori T, et al. Long-term outcome after drug-eluting stents implantation: target lesion versus nontarget lesion repeated intervention. Int J Cardiol 2010;145:322-4.

9 American Heart Journal Volume 167, Number 3 Chin et al 400.e1 Appendix. Variables in the multivariable model for procedural success and in-hospital mortality Age Sex Race Body mass index ST-segment elevation myocardial infarction on presentation Cardiogenic shock on presentation History of congestive heart failure History of previous valve surgery History of cerebrovascular disease History of peripheral vascular disease History of chronic lung disease Intra-aortic balloon pump placed before PCI Left ventricular ejection fraction Highest risk lesion Pre-procedural TIMI flow grade New onset diabetes mellitus Glomerular filtration rate On dialysis New York Heart Association Class heart failure on presentation Coronary segment Urgent versus emergent versus salvage PCI status