Original Contribution

Outcomes of Culprit Versus Multivessel PCI in Patients With Multivessel Coronary Artery Disease Presenting With ST-Elevation Myocardial Infarction Complicated by Shock

Matthew A. Cavender, MD1, Jeevanantham Rajeswaran, PhD1, Linda DiPaola, BA1,  Penny Houghtaling, MS1, Michael S. Kiernan, MD2, Andrew N. Rassi, MD1, Venu Menon, MD1, Patrick W. Whitlow, MD1, Stephen G. Ellis, MD1, Mehdi H. Shishehbor, DO, MPH1

Matthew A. Cavender, MD1, Jeevanantham Rajeswaran, PhD1, Linda DiPaola, BA1,  Penny Houghtaling, MS1, Michael S. Kiernan, MD2, Andrew N. Rassi, MD1, Venu Menon, MD1, Patrick W. Whitlow, MD1, Stephen G. Ellis, MD1, Mehdi H. Shishehbor, DO, MPH1

Download a PDF of this article.

Abstract: Background. The optimal revascularization strategy in patients with multivessel coronary artery disease (MVCAD) who present with ST-elevation myocardial infarction (STEMI) and shock is undefined. We aimed to determine differences in survival among patients with MVCAD presenting with STEMI complicated by shock treated with percutaneous coronary intervention (PCI) of the infarct-related artery alone (culprit-only PCI) versus multivessel PCI (MVPCI). Methods. Patients with MVCAD and STEMI complicated by shock who underwent PCI between January 1, 2002 and May 31, 2010 were identified (n = 199). Differences in survival between patients undergoing culprit-only PCI versus MVPCI were assessed using a multiphase survival model and propensity matching. Results. MVPCI was used in 22% of patients (n = 43). Patient characteristics were similar in the groups, although more patients treated with MVPCI met the National Cardiovascular Data Registry definition of shock. Death was higher in patients treated with MVPCI at 1 month (27% vs 46%) and 8 years (65% vs 75%; P=.04). The early risk of death was higher in the patients treated with MVPCI when compared to patients treated with culprit-only PCI (coefficient: 0.66 ± 0.25; P=.009), but not the late risk of death (coefficient: -0.18 ± 0.58; P=.70). However, in a propensity-matched cohort (n = 64), there were no differences in the risk of death over the early (coefficient: 0.50 ± 0.37; P=.16) or late phase (P>.90). Conclusion. Patients undergoing MVPCI for STEMI-related shock are clinically different than those treated with culprit PCI only; however, after risk adjustment both groups have similar short- and long-term outcomes. Prospective studies are needed to determine the optimal revascularization strategy in this high-risk population.

J INVASIVE CARDIOL 2013;25(5):218-224

Key words: percutaneous coronary intervention, acute myocardial infarction, cardiogenic shock



ACC = American College of Cardiology
AHA = American Heart Association
CAD = coronary artery disease
CABG = coronary artery bypass grafting
CCF = Cleveland Clinic Foundation
COPD = chronic obstructive pulmonary disease
VA = cerebral vascular accident
CVD = cerebral vascular disease
IABP = intra-aortic balloon pump
LVEF = left ventricular ejection fraction
MI = myocardial infarction
MVCAD = multivessel coronary artery disease
NCDR = National Cardiovascular Disease Registry
NYHA = New York Heart Association
PAD = peripheral arterial disease
SHOCK = Should We Emergently Revascularize Occluded Coronaries for Cardiogenic Shock
SSDI = Social Security death index
STEMI = ST-elevation myocardial infarction
PCI = percutaneous coronary intervention
SysBP = systolic blood pressure

Prior data from the Should We Emergently Revascularize Occluded Coronaries for Cardiogenic Shock (SHOCK) trial and registry have shown that prompt revascularization can improve outcomes among patients with ST-elevation myocardial infarction (STEMI) complicated by cardiogenic shock.1 While multivessel coronary artery disease (MVCAD) is common in this population, with some series suggesting an incidence of >50%, the optimal revascularization strategy remains controversial.2-5 Current revascularization strategies for patients with MVCAD include percutaneous coronary intervention (PCI) for the infarct-related artery only (culprit-PCI) followed by either medical therapy or staged PCI of the non-infarct related artery, percutaneous revascularization of both infarct and non-infarct related arteries (MVPCI), or coronary artery bypass graft surgery.

There are few data to suggest which treatment strategy is associated with the best outcomes, since prior studies have had significant limitations. An analysis from the CathPCI Registry found MVPCI to be associated with worse outcomes in patients with cardiogenic shock and MVCAD even after adjusting for medical comorbidities and acuity; however, this study was limited to in-hospital outcomes.6

There have been no studies evaluating the long-term outcomes associated with culprit-PCI compared to MVPCI. This is especially important in light of data that have shown that patients with cardiogenic shock who survived their hospital stay have a good long-term prognosis.7 Therefore, we sought to determine the frequency of culprit-PCI and MVPCI, identify differences in the populations treated with culprit-PCI as compared to MVPCI, understand the predictors of multivessel revascularization at the time of presentation, and evaluate differences in both short- and long-term survival seen between the different revascularization strategies in patients with MVCAD who present with STEMI complicated by shock.


Patient population, definitions, and outcomes. All patients undergoing revascularization with PCI for STEMI at a tertiary care hospital between January 1, 2002 and May 31, 2010 were identified (n = 1825). In order to study a cohort of patients undergoing revascularization for STEMI with MVCAD and cardiogenic shock, patients with single-vessel coronary artery disease (SVCAD), no evidence of cardiogenic shock, and patients with definite indications for surgery (such as significant valvular heart disease or mechanical complications of the myocardial infarction) were excluded from the analysis (Figure 1).

MVCAD was defined as at least 50% stenosis in two major epicardial coronary arteries.8 The extent of CAD was graded by the visual consensus of two observers at the time of catheterization. Coronary disease was classified and described using the number of diseased vessels with 50% stenosis in 2 or 3 major epicardial vessels and the Duke CAD Severity Index.9,10 The Duke CAD Index is a well-validated score that measures the overall burden of CAD. The index takes into account the artery, location within the artery, severity, and number of diseased vessels. The index has been shown to have prognostic implications for patients with CAD.11,12 Severe CAD was considered to be present in patients with a Duke CAD index >42.12

Cardiogenic shock was defined as patients who had a systolic blood pressure (sysBP) < 90 mm Hg, needed an intra-aortic balloon pump (IABP) for hemodynamic support, or met the standardized CathPCI Registry definition (sustained episode of sysBP <90 mm Hg, and/or cardiac index <2.2 L/min/m2, and/or parenteral inotropic or vasopressor agents or mechanical support [eg, IABP, extracorporeal circulation, ventricular assist devices] needed to maintain blood pressure and cardiac index above those specified levels).13 All patients who were classified as having shock based only on placement of IABP were reviewed and excluded if the IABP was placed for reasons other than cardiogenic shock.

Patients were divided into two cohorts based upon revascularization strategy pursued at the time of presentation. Patients were considered to undergo culprit-PCI if they underwent isolated PCI of the infarct-related artery only (including the left main artery). Patients were considered to have MVPCI if they underwent revascularization of another major epicardial coronary artery in addition to PCI of the infarct-related artery. In patients undergoing culprit-PCI, the infarct-related artery was considered to be the artery that underwent revascularization. The infarct-related artery was denoted in patients undergoing multivessel PCI; however, some patients had more than 1 infarct-related artery identified. In those cases, retrospective chart review was performed to determine the infarct-related artery based upon electrocardiographic and angiographic findings at the time of presentation.

The primary outcome of this study was all-cause mortality as determined using the Social Security Death Index (SSDI). Those patients not found in the SSDI were censored at the time of last patient contact. A total of 517 patient years were available for analysis.

Statistical analysis. Continuous variables are presented as mean ± standard deviation. Tests of significance were performed using Wilcoxon rank-sum nonparametric tests. Categorical data are described using frequencies and percentages with comparisons made using χ2 tests. Fisher’s exact test was used when the frequency was less than 5. Some baseline characteristics and procedural variables examined in multivariable analyses had missing values. A 5-fold multiple imputation using a Markov Chain Monte Carlo technique was used to impute these missing values.14

Survival models. Survival was first assessed non-parametrically using the Kaplan-Meier and the log-rank test for statistical significance. Survival was further assessed parametrically using a multiphase hazard model. A multiphase hazard model was used since the baseline hazard for death was high at the time of presentation but decreased over time suggesting that different risk factors may influence early and late risk.15,16 The parametric model was used to resolve these different phases of instantaneous risk of death (hazard function) and to estimate shaping parameters. The estimated regression coefficients and their variance-covariance matrix were determined for each imputed complete dataset. These estimates were then combined to yield final regression coefficient estimates, the variance-covariance matrix, and P-values.14 The results of the multiphase hazard model are presented with their coefficients rather than the hazard ratio because some continuous variables were transformed to meet statistical model assumptions, making the hazard ratio difficult to interpret. The results from the logistic regression model are also presented with their coefficients for similar reasons. Uncertainty is expressed by 95% confidence limits.

Using baseline demographic and variables available prior to PCI (Appendix 1), a multivariable logistic regression model identifying the factors associated with MVPCI group was developed.17,18 Variable selection was performed using a bootstrap bagging (bootstrap aggregation) technique that retained variables in the model with a P-value of .05. This was done by randomly selecting a patient from the original dataset and beginning a new dataset. The original dataset continued to be sampled until the new dataset was 100% the size of the original. Second, risk factors were identified in the new data using automated forward stepwise selection and stored. These steps were then repeated 500 times. Finally, the frequency of occurrence of variables related to group membership was ascertained, which indicated the reliability of each variable (aggregation step). All variables with bootstrap reliability of 50% or greater were retained in the guided analysis. 

Having established a parsimonious model, we added other variables representing patient demographics, symptoms, and comorbidities that might be related to unrecorded selection factors (saturated model) (Appendix 1). 

Propensity analysis. In order to control for potential confounders, a propensity score was calculated for each patient by solving the saturated model for the probability of being in the MVPCI group. Then, using only the propensity score, culprit-PCI group cases were matched to MVPCI cases using a greedy matching strategy.19 Culprit-PCI cases whose propensity scores deviated more than 0.10 from those of MVPCI cases were considered unmatched.20 The analysis performed on the entire cohort was then repeated using only those patients that could be matched on the basis of their propensity for receiving MVPCI. In addition, a multivariable analysis model for the overall cohort that controlled for the propensity to receive multivessel PCI was developed to validate/confirm the findings of the propensity-matched analyses.

The institutional review board at the Cleveland Clinic approved this analysis. All tests of significance were two-sided and a P-value of <.05 was considered significant. Statistical analysis was performed using SAS version 9.2 (SAS Institute).


Patients with MVCAD presenting with STEMI complicated by cardiogenic shock were identified (n = 199) from within the overall cohort of patients with STEMI undergoing PCI. The mean follow-up for the entire cohort was 2.6 years. The majority of these patients (n = 156; 78%) were treated with culprit-PCI of the infarct-related artery alone, while 22% (n = 43) were treated with MVPCI (Table 1).

Baseline characteristics were similar in the two groups; however, MVPCI was associated with lower ejection fraction. The proportion of patients with either a sysBP <90 mm Hg or IABP were similar between the two groups. The most common infarct-related artery was the left anterior descending coronary artery, and there were no differences seen in either the infarct-related artery or extent of CAD as measured by the Duke CAD Index (Table 2).

Unadjusted comparison. Patients treated with MVPCI with higher overall incidence of death using Kaplan-Meier estimations both at 1 month (27% vs 46%) and 8 years (65% vs 75%; P=.04). The risk of death for the overall cohort was highest in the early phase and plateaued over the late phase. While the early risk of death was high in both groups, the early risk of death was considerably higher in the patients treated with MVPCI when compared to patients treated with culprit-PCI (coefficient: 0.66 ± 0.25; P=.009) (Figure 2A).

The differences seen in the risk of death in the early phase were attenuated during the late phase such that there was no difference in the risk of death between the groups in that phase of the model (P=.70) (Figure 2B).

Propensity analysis.Given the concern about baseline differences in the groups, a separate analysis was performed using a cohort matched based upon their propensity to receive MVPCI (Figure 3).

The strongest predictors of receiving MVPCI were clinical factors, such as increased heart rate, lower LVEF and meeting the CathPCI Registry definition for cardiogenic shock (Table 3). A proportion of patients undergoing MVPCI (n = 32) were matched with patients treated with culprit-PCI who were considered to have a similar propensity for receiving MVPCI (n = 32). Baseline and procedural characteristics between the matched groups were similar.

Propensity-adjusted comparison. In the propensity-matched cohort, there was no difference in the risk of early death (coefficient: 0.50 ± 0.37; P=.16) or late death (P=.90) in patients treated with MVPCI when compared to patients treated with culprit-PCI (Figures 4A and 4B).

A survival model for the overall cohort that controlled for the propensity to receive MVPCI was developed to validate/confirm the findings of the propensity-matched cohort. Incremental risk factors for death in the early hazard phase were older age, LVEF <30%, higher heart rate, clinical shock, and longer lesion length; however, MVPCI was not a predictor for either early death (P=.13) or late death (P=.24).


Our analysis found that, among patients with MVCAD who present with STEMI complicated by shock, MVPCI is common and associated with a higher early hazard of death as compared to those patients undergoing culprit-PCI. The early differences in the hazard of death were not present during the late phase, suggesting that differences in the initial presentation drive the differences in outcomes seen between the two groups. The early increased hazard seen with MVPCI was not seen after propensity adjustment, implying that even among patients with cardiogenic shock outcomes are driven by differences in patient factors and the acuity of illness rather than the extent of revascularization during the initial presentation. We also found the long-term outcomes among those who survive the initial infarction to be similar irrespective of the extent of initial revascularization. These results are consistent with previously published data showing similar long-term outcomes among patients presenting with STEMI with and without cardiogenic shock if they survive the initial hospitalization.7,21

Current American College of Cardiology (ACC) and American Heart Association (AHA) guidelines state that PCI of the non-infarct related artery should not be performed at the time of PCI for patients not in cardiogenic shock. While not given a specific recommendation, the guidelines do note that in patients with cardiogenic shock, revascularization of a non-infarct related artery could be considered if the lesion “appeared to be flow limiting in patients with hemodynamic instability” or the patient was having ongoing symptoms.22 The European Society of Cardiology guidelines for the use of PCI offer stronger support for multivessel revascularization, stating, “complete revascularization has been recommended with PCI performed in all critically stenosed large epicardial coronary arteries.”23 There are few available data on which to make these suggestions, since the majority of studies evaluating multivessel revascularization in patients with STEMI have not subdivided patients based on the presence of cardiogenic shock; however, these recommendations are based in part on the hypothesis that complete revascularization would reduce ischemia and improve cardiac function. The largest study to date of patients in cardiogenic shock came from the CathPCI Registry database. Among the 3087 patients identified with cardiogenic shock, multivessel revascularization was associated with higher odds of in-hospital death even after adjusting for potential confounders.6

Our study is the first to compare long-term outcomes of culprit-PCI versus MVPCI in patients with MVCAD who present with STEMI complicated by cardiogenic shock. Individuals treated with MVPCI appear to be more acutely ill at the time of presentation; however, their long-term outcomes were similar to those who underwent culprit-PCI only. It is possible that MVPCI could have modified the long-term risk of these patients; more importantly, it is clear that MVPCI was not associated with harm. These data provide support for the current ACC/AHA guidelines and suggest that interventional cardiologists can use their judgment during primary PCI in patients with cardiogenic shock and pursue revascularization of lesions they deem to be contributing to ongoing symptoms or hemodynamic compromise. 

Study limitations. Our findings should be considered within the context of the data from which they were derived and the limitations of our analysis.  Our data are representative of a single tertiary-care center with operators who have considerable experience in primary PCI and the treatment of acute coronary syndromes. In addition, it is retrospective in nature and includes a relatively small number of patients; however, we used robust statistical methods including multiphase survival models with bootstrapping techniques that allowed for the control of multiple patient characteristics including the severity of CAD. While it is possible that with an increased number of patients, differences in the two groups may have been recognizable, our preliminary analyses suggest that it would have taken approximately 500 additional patients for the current differences to be statistically significant (assuming that the current trends continued). While this does not seem to be a significant number, it is worth noting that the SHOCK trial had only 82 patients in the revascularization arm. The small number of patients in studies of cardiogenic shock illustrates the challenges of studying this unique and high-risk population. 


Our data show that multivessel revascularization is common in patients with MVCAD who present with STEMI complicated by cardiogenic shock. Patients undergoing MVPCI have long-term outcomes comparable to those of patients treated with PCI of the infarct-related artery alone. These data provide the first long-term study supporting the ACC/AHA guidelines and show that multivessel interventions can be considered in select patients without fear that more extensive revascularization worsens outcomes. Furthermore, these data highlight the impact of both measured and unmeasured confounders on the outcomes of patients undergoing different treatment strategies for STEMI. As a result, randomized studies are needed to definitively determine the best method of revascularization in this unique population.

Appendix 1. Candidate variables for logistic regression model determining predictors of multivessel PCI. Asterisk (*) denotes variables included in the propensity model.

Gender*; age (in years)*; weight; height; body surface area; body mass index*; race (white, black)*; New York Heart Association functional class*; prior myocardial infarction*; left ventricular ejection fraction*; history of coronary artery bypass graft surgery*; history of PCI; systolic blood pressure*; heart rate*; shock based on the CathPCI Registry definition*; polyvascular disease*; diabetes*; hypertension*; peripheral artery disease*; hyperlipidemia*; chronic obstructive pulmonary disease*; prior/ongoing tobacco abuse*; renal disease*; coronary anatomy (left anterior descending, left circumflex, left main, right coronary artery 50% stenosis); infarct-related artery (mid-left anterior descending*, proximal left anterior descending*, left circumflex*, left main*, right coronary artery*); length of longest lesion treated (in centimeters)*; reference vessel diameter*; Duke coronary artery disease index*; intra-aortic balloon pump placed prior to PCI*; date of PCI*.


  1. Hochman JS, Sleeper LA, White HD, et al. One-year survival following early revascularization for cardiogenic shock. JAMA. 2001;285(2):190-192.
  2. Widimsky P, Holmes DR Jr. How to treat patients with ST-elevation acute myocardial infarction and multivessel disease? Eur Heart J. 2011;32(4):396-403.
  3. Sorajja P, Gersh BJ, Cox DA, et al. Impact of multivessel disease on reperfusion success and clinical outcomes in patients undergoing primary percutaneous coronary intervention for acute myocardial infarction. Eur Heart J. 2007;28(14):1709-1716.
  4. Abbott JD, Ahmed HN, Vlachos HA, Selzer F, Williams DO. Comparison of outcome in patients with ST-elevation versus non-ST-elevation acute myocardial infarction treated with percutaneous coronary intervention (from the National Heart, Lung, and Blood Institute Dynamic Registry). Am J Cardiol. 2007;100(2):190-195.
  5. Varani E, Balducelli M, Aquilina M, et al. Single or multivessel percutaneous coronary intervention in ST-elevation myocardial infarction patients. Catheter Cardiovasc Interv. 2008;72(7):927-933.
  6. Cavender MA, Milford-Beland S, Roe MT, Peterson ED, Weintraub WS, Rao SV. Prevalence, predictors, and in-hospital outcomes of non-infarct artery intervention during primary percutaneous coronary intervention for ST-segment elevation myocardial infarction (from the National Cardiovascular Data Registry). Am J Cardiol. 2009;104(4):507-513.
  7. Hochman JS, Sleeper LA, Webb JG, et al. Early revascularization and long-term survival in cardiogenic shock complicating acute myocardial infarction. JAMA. 2006;295(21):2511-2515.
  8. Serruys PW, Morice MC, Kappetein AP, et al. Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease. N Engl J Med. 2009;360(10):961-972.
  9. Bhatt DL, Peterson ED, Harrington RA, et al. Prior polyvascular disease: risk factor for adverse ischaemic outcomes in acute coronary syndromes. Eur Heart J. 2009;30(10):1195-1202.
  10. Mark DB, Nelson CL, Califf RM, et al. Continuing evolution of therapy for coronary artery disease. Initial results from the era of coronary angioplasty. Circulation. 1994;89(5):2015-2025.
  11. Felker GM, Shaw LK, O’Connor CM. A standardized definition of ischemic cardiomyopathy for use in clinical research. J Am Coll Cardiol. 2002;39(2):210-218.
  12. Cavender MA, Alexander KP, Broderick S, et al. Long-term morbidity and mortality among medically managed patients with angina and multivessel coronary artery disease. Am Heart J. 2009;158(6):933-940.
  13. Standardized Data Definitions and Elements in the NCDR. Accessed May 25, 2012. Available at http://www.ncdr.com/WebNCDR/ELEMENTS.ASPX.
  14. Rubin DB. Multiple Imputation for Non-Response in Surveys. 1st edition. New York: Wiley; 1997.
  15. Blackstone EH, Naftel DC, Turner ME. The decomposition of time-varying hazard into phases, each incorporating a separate stream of concomitant information. J Am Stat Assoc. 1986;81:615-624.
  16. Kaplan EL Meier P. Nonparametric Estimation from Incomplete Observations. J Am Stat Assoc. 1958;53:457-481.
  17. Breiman L. Bagging predictors. Machine Learning. 1996;24:123-140.
  18. Blackstone EH. Breaking down barriers: helpful breakthrough statistical methods you need to understand better. J Thorac Cardiovasc Surg. 2001;122(3):430-439.
  19. Bergstralh EJ, Konsanke JL. Computerized matching of cases to controls. Technical report #56. Mayo Foundation, 1995.
  20. Blackstone EH. Comparing apples and oranges. J Thorac Cardiovasc Surg. 2002;123(1):8-15.
  21. Singh M, White J, Hasdai D, et al. Long-term outcome and its predictors among patients with ST-segment elevation myocardial infarction complicated by shock: insights from the GUSTO-I trial. J Am Coll Cardiol. 2007;50(18):1752-1758.
  22. Levine GN, Bates ER, Blankenship JC, et al. 2011 ACCF/AHA/SCAI guideline for percutaneous coronary intervention: executive summary: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines and the Society for Cardiovascular Angiography and Interventions. Circulation. 2011;124(23):2574-2609.
  23. Wijns W, Kolh P, Danchin N, et al. Guidelines on myocardial revascularization. Eur Heart J. 2010;31(20):2501-2555.


From the 1Heart and Vascular Institute, Cleveland Clinic, Cleveland, Ohio and the 2Division of Cardiology, Tufts University, Boston, Massachusetts.

Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Kiernan is on the Advisory Board for Gambro. The remaining authors report no disclosures.

Manuscript submitted November 19, 2012 and accepted February 6, 2013.

Address for correspondence: Dr Mehdi H. Shishehbor, Department of Cardiovascular Medicine, Cleveland Clinic, 9500 Euclid Avenue, J3-5, Cleveland, OH 44195. Email: shishem@ccf.org