Author + information
- Received July 9, 2008
- Accepted August 5, 2008
- Published online October 1, 2008.
- Mariusz Kruk, MD, PhD⁎,
- Jacek Kadziela, MD, PhD⁎,
- Harmony R. Reynolds, MD, FACC†,
- Sandra A. Forman, MA‡,
- Zygmunt Sadowski, MD, FACC, FESC⁎,
- Bruce A. Barton, PhD‡,
- Daniel B. Mark, MD, FACC§,
- Aldo P. Maggioni, MD, FESC∥,
- Jonathan Leor, MD, FACC¶,
- John G. Webb, MD, FACC#,
- Michael Kapeliovich, MD, EC⁎⁎,
- Jose A. Marin-Neto, MD, PhD, FACC††,
- Harvey D. White, MD, FACC‡‡,
- Gervasio A. Lamas, MD, FACC§§ and
- Judith S. Hochman, MD, FACC†,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Judith S. Hochman, New York University School of Medicine; 530 First Avenue, Skirball 9R, New York, New York 10016
Objectives This study sought to determine predictors of outcome and examine the influence of baseline risk on therapeutic impact of late mechanical opening of a persistently occluded infarct related artery after myocardial infarction in stable patients.
Background Previous studies in patients with acute coronary syndromes suggest that the impact of infarct-related artery recanalization on clinical outcome is greatest in patients at highest risk.
Methods Of 2,201 patients (age 58.6 ± 11.0 years) with infarct-related artery occlusion on days 3 to 28 after myocardial infarction in the OAT (Occluded Artery Trial) study, 1,101 were assigned to percutaneous coronary intervention (PCI) and 1,100 to medical therapy alone and followed for a mean of 3.2 years. The primary end point was a composite of death, reinfarction, or New York Heart Association functional class IV heart failure. Interaction of treatment effect with tertiles of predicted survival were examined using the Cox survival model.
Results The 5-year rate for the primary end point was 18.9% versus 16.1% for patients assigned to PCI and medical treatment alone, respectively (hazard ratio [HR]: 1.14, 95% confidence interval [CI]: 0.92 to 1.43, p = 0.23). Lack of benefit of PCI was consistent across the risk spectrum for both the primary end point and total mortality, including for the highest tertile (33.9% PCI vs. 27.3% medical treatment alone, HR: 1.27, 99% CI: 0.87 to 1.85 primary end point and 23.5% PCI vs. 21.7% medical treatment alone, HR: 1.16, 99% CI: 0.73 to 1.85 mortality). The independent predictors of the composite outcome were history of heart failure (HR: 2.06, p < 0.001), peripheral vascular disease (HR: 1.93, p = 0.001), diabetes (HR: 1.49, p = 0.002), rales (HR: 1.88, p < 0.001), decreasing ejection fraction (HR: 1.48 per 10%, p < 0.001), decreasing days from myocardial infarction to randomization (HR: 1.04 per day, p < 0.001), and decreasing glomerular filtration rate (HR: 1.11 per 10 ml/min/1.73 m2, p < 0.001).
Conclusions In the OAT study, there was no variation in the effect of PCI on clinical outcomes at different levels of patient risk, including the subset with very high event rates. (Occluded Artery Trial [OAT]; NCT00004562)
Total occlusion of the infarct-related artery (IRA) may be detected in up to one-third of all patients surviving the acute phase of myocardial infarction (MI). Such patients typically have not received acute reperfusion therapy due to presentation outside the time window of demonstrated survival benefit, and there are a paucity of data regarding prognostic factors in this population (1–3). Recently, the OAT (Occluded Artery Trial) study has shown that contemporary percutaneous methods used for mechanical recanalization of IRA performed in stable patients with 1- or 2-vessel coronary disease on days 3 to 28 following MI do not result in better prognosis when compared with optimal medical treatment (3). Previous clinical experience with other subsets of acute coronary syndrome (ACS) patients, including cardiogenic shock after MI, suggests that more aggressive, interventional therapies are most beneficial in highest risk patients (4–7). To test this, we determined predictors of outcome in the OAT study and examined the influence of baseline risk on the therapeutic impact of mechanical opening of persistent IRA and tested the hypothesis that those at greatest risk may derive benefit.
The design and methods of OAT have been described in detail (3,8). In summary, 2,201 patients were enrolled (2,166 between February 2000 and December 2005 in the main OAT study (3) and 35 patients in the extension phase of the OAT-NUC (Viability and Remodeling in OAT Ancillary Study) substudy in 2006) and observed for a mean of 3.2 years. Patients who had total occlusion of the IRA on calendar days 3 to 28 (minimum 24 h) after MI symptom onset were eligible if they were high risk (i.e., ejection fraction [EF] <50% or proximal occlusion of the IRA subtending ≥25% left ventricular [LV] myocardium) and remained in stable clinical condition. Major exclusion criteria included New York Heart Association (NYHA) functional class III or IV heart failure, shock, a serum creatinine concentration >2.5 mg/dl (221 μmol/l), angiographically significant left main or 3-vessel coronary artery disease, angina at rest, and severe ischemia on stress testing (not routinely required but warranted if the infarct zone was not akinetic or dyskinetic or multivessel disease was present).
Eligible patients were randomly assigned to percutaneous coronary intervention (PCI) with stent placement and optimal medical therapy (n = 1,101) or optimal medical therapy alone (n = 1,100). Per protocol, all patients were to receive optimal medical therapy, including aspirin, anticoagulation if clinically indicated, angiotensin-converting enzyme inhibition or blockade of angiotensin II receptor, beta-blockade, and lipid-lowering therapy, unless contraindicated. Thienopyridine therapy was to be initiated before PCI and continued for at least 2 to 4 weeks in patients who underwent stenting and was recommended for 6 to 12 months in all patients.
Images from the qualifying angiogram and PCI were reviewed at the angiography core laboratory. Cardiac markers (preferably creatine kinase myocardial band or, if not available, troponin I or T or creatine kinase) were to be measured routinely in both groups 3 times during the first 48 h after randomization and within 24 h after PCI in patients assigned to PCI.
Institutional review boards at the participating centers approved the study protocol, and all patients provided written informed consent. Funding for the present work was provided by a grant from the National Heart, Lung, and Blood Institute.
Study end points
The primary end point of the OAT study was a composite of death from any cause, reinfarction, or NYHA functional class IV heart failure with hospitalization or admission in a short-stay unit. This cumulative composite end point and death alone assessed at 5 years comprised the outcomes of the current analysis.
The study end point events were adjudicated by an independent mortality and morbidity classification committee, which was unaware of the treatment assignments. The definition of reinfarction, which required 2 out of 3 criteria, has been provided previously (3).
Baseline characteristics that had <10% missing data were used for this analysis (Table 1). Data were summarized as frequencies and percentages for categorical variables and as the median (with interquartile range) for continuous variables and compared within the risk strata using the chi-square test/Fisher exact test or analysis of variance, respectively. Univariable and multivariable Cox proportional hazards models were developed to evaluate the relation between the baseline characteristics and the occurrence of the composite of death, myocardial reinfarction, or NYHA functional class IV heart failure during follow-up, and, separately, death alone. Factors that were expected to affect risk and likely treatment interaction were prespecified by the OAT protocol (8). However, to determine independent predictors of outcome, all baseline variables entered the multivariable stage, irrespective of the results of univariable analyses. Missing values were imputed as means for continuous variables and, equivalently, medians for binary variables. Backward elimination was used with p ≤ 0.01 for variables remaining in the final model. Secondary analysis included the addition of discharge medications to the models based on the baseline data. The prognostic discriminatory capacity of the models based on the baseline variables was measured with the C statistic representing the area under the receiver operating characteristic curves for prediction of the composite end point and of death alone. Calibration of the OAT risk multivariable models (i.e., how closely predicted outcomes agreed with actual outcomes) was evaluated with Hosmer-Lemeshow chi-square statistics for goodness of fit.
Prior to the analysis, we prespecified that the patients would be divided into tertiles of risk based on the multivariable Cox model predictive of the primary end point. Kaplan-Meier curves for patients within consecutive risk tertiles were presented graphically and compared with log-rank tests. Interaction between study assigned treatment and the risk strata was assessed with the use of a Cox regression model.
To control for the type I error rate, it was prespecified by the study protocol that a p value of <0.01 would be considered as showing evidence of differences in secondary analyses (8). SAS software version 9.1.3 (SAS Institute, Cary, North Carolina) was used for statistical analyses.
The cumulative 5-year composite primary end point of death, myocardial reinfarction, or NYHA functional class IV heart failure was 17.5%; the mortality rate was 11.7%. No significant differences were observed between patients assigned to PCI and patients assigned to optimal medical therapy alone for the composite outcome (18.9% vs. 16.1%, respectively; hazard ratio [HR]: 1.14, 95% confidence interval [CI]: 0.92 to 1.43, p = 0.23) or for death (11.7% vs. 11.8%, respectively; HR: 1.00, 99% CI: 0.69 to 1.47, p = 0.99). The 1-year composite primary end point was 9.2%, and death was 4.3%. Univariable analysis of predictors of the composite outcome and death are listed in Table 1.
The independent predictors of the primary composite end point included history of heart failure prior to the index MI (HR: 2.06, p < 0.001), peripheral vascular disease (HR: 1.93, p = 0.001), diabetes (HR: 1.49, p = 0.002), rales at the time of randomization (HR: 1.88, p < 0.001), EF (HR: 1.48 per 10% decrease, p < 0.001), days from MI to randomization (HR: 1.04 per day decrease, p < 0.001), and reduced glomerular filtration rate (HR: 1.11 per 10 ml/min/1.73 m2, p < 0.001). Independent predictors of death were: Killip class >1 during index MI (HR: 1.75, p < 0.001), history of cerebrovascular disease (HR: 2.28, p = 0.002), angina (HR: 1.57, p = 0.005), history of heart failure (HR: 2.11, p = 0.004), EF (HR: 1.51 per 10% decrease, p < 0.001), days from MI to randomization (HR: 1.03 per day decrease, p = 0.008), and reduced glomerular filtration rate (HR: 1.18, per 10 ml/min/1.73 m2, p < 0.001) (Figs. 1A and 1B, respectively).
Both multivariable Cox models were modified by addition of discharge medications. The primary outcome was further predicted by digoxin treatment (HR: 2.07, p = 0.001) in addition to each of the variables noted in the first model except for rales, whereas death was additionally predicted by insulin prescription at discharge (HR: 1.85, p = 0.006) but not by Killip class >1. Multivariable Cox models were not modified by the addition of the study treatment (PCI vs. medical) assignment.
Prognostic capacity and calibration of the multivariable Cox regression models
The areas under the receiver operating curves for the Cox models in prediction of both the primary outcome and death showed good discrimination (C statistics = 0.720 for the composite end point, and C statistics = 0.731 for death). The Hosmer-Lemeshow statistics for the respective OAT multivariable Cox models were not significant, indicating little departure from a perfect fit (chi-square = 7.710 and p = 0.462) for the composite outcome and (chi-square = 6.09 and p = 0.638) for death.
Based on the multivariable Cox model for the primary end point, we divided the study patients into 3 equal groups based on increasing risk profile. The cumulative 5-year life table event rates for primary end point for patients within low-, medium-, and high-risk tertiles were 8.6%, 13.4%, and 30.6%, respectively. Accordingly, death occurred in 5.5%, 6.9%, and 22.6% of cases by tertile.
Baseline characteristics of the 2,201 study patients divided based on risk tertiles is presented in Table 2. Increasing risk tertiles were associated with older age, female gender, more comorbidities, diabetes, obesity, heart failure prior to study entry, higher heart rate, lower diastolic blood pressure, increasing glucose levels, and decreasing glomerular filtration rate. Higher risk patients more often had ST-segment elevation during the index MI, and the time from the qualifying event to randomization was substantially shorter than in the lower risk patients. Other features defining patients with increasing risk tertiles were higher prevalence of the left anterior descending artery as the culprit vessel, lower likelihood of collaterals, more extensive coronary artery disease, and lower EF.
No significant differences were observed between the risk subgroups with respect to the periprocedural composite event. However, periprocedural death occurred more often in the highest risk group of patients (n = 5) than in the medium- (n = 1) or low-risk (n = 0) groups (p = 0.01, Fisher exact test).
Interaction of treatment effect with the risk strata
The survival curves depicting the occurrence of a composite end point and mortality alone within risk tertiles are provided in Figure 2, and the event rates within consecutive risk tertiles for the study treatment groups (PCI vs. medical) are presented in Table 3. There were no statistically significant differences between the 2 treatment groups for any risk tertile, but in the highest risk patients there was a weak trend toward more events (composite end point) in the PCI group (log-rank p = 0.10).
There was no significant interaction between the study treatment and the risk tertile subgroups with respect to either the composite end point (interaction p value = 0.51) or death (interaction p value = 0.17) during follow-up.
Our study extends the primary findings of the OAT study by determining clinical variables independently associated with long-term adverse prognosis and indicating a consistent lack of benefit of late mechanical recanalization on clinical outcomes irrespective of baseline risk. Importantly, our analysis included over 700 patients with a high cardiovascular event rate (5-year composite outcome rate of 30.6% with death in 22.6%). It has been hypothesized that patients at higher risk of events would have the most to gain from PCI of the IRA; this is not the case based on the present analysis. In fact, we identified a trend toward worse clinical outcomes including higher PCI procedural risk in the OAT study within the highest risk tertile.
The late open artery hypothesis included several theoretical cardiovascular benefits of late opening of the IRA: prevention of infarct expansion, improvement in LV function by recovery of hibernating myocardium, increased electrical stability, or capacity of a patent IRA to provide collaterals in the event of future remote occlusion. Despite 83% patency rate for intervened IRA after 1 year and an apparent positive impact of IRA recanalization on LV remodeling reported in the TOSCA (Total Occlusion Study of Canada)-2 trial, the results of the OAT study have shown that these potential benefits do not translate into lower rates of heart failure, death, and reinfarction during 5 years of follow up.
According to our data, the 5-year prognosis of the OAT study patients seemed to be determined primarily by comorbidities, and the cardiac component was represented only modestly by left ventricular ejection fraction (LVEF) or heart failure. In the angiographic OAT substudy, TOSCA-2, PCI of the IRA had no influence on LVEF (9) at 1 year. Lack of influence of IRA recanalization on LVEF may partially explain the lack of interaction between the risk strata and the study treatment observed in the current analysis. Alternatively, the potential effect of EF may have been moderated by use of optimal medical therapy, which is particularly effective for risk reduction among patients with low LVEF.
The OAT study population was composed primarily of patients with 1- or 2-vessel disease, and the comorbid conditions identified as predictors of outcome are likely to act at least in part via acceleration of coronary artery disease progression. Predictors of outcome in the OAT study are largely concordant with predictors of outcome in prior studies of stable coronary disease and ACS patients. The risk in ACS is largely determined by pump function, which is typically reflected by heart rate, blood pressure, or Killip class at presentation (10–15). The LVEF consistently correlated to risk of events, particularly death and heart failure, in both acute and chronic coronary artery disease (11,14,15). Renal dysfunction and diabetes have each been shown to be strong risk determinants of cardiovascular events in multiple previous studies as well as in our results (16–18). Peripheral vascular and cerebrovascular diseases constitute important cardiovascular risk determinants in patients with and without established coronary disease possibly related to more severe coronary atherosclerosis (19–23). The impact of peripheral vascular disease, nearly doubling the risk of the composite event observed in our study, was comparable to previously reported data regarding patients undergoing PCI (20). Additionally, history of stroke may be a common denominator of patients with atrial fibrillation or more advanced heart failure (19,23).
The addition of discharge medications to the multivariable Cox models showed that digoxin use was independently associated with the composite outcome and that insulin treatment was associated with death during follow-up. These results likely reflect the association between use of these medications and the presence of more advanced heart failure or unaccounted comorbidities such as atrial fibrillation in the case of digoxin. Similarly, insulin treatment may be a marker of a more advanced stage or longer duration of diabetes (16,18). However, we cannot exclude causal relations. The lack of an apparent protective effect of angiotensin-converting enzyme or angiotensin-receptor blocker or beta-blocker is likely due to limited power related to the fact that most patients received these medications.
Increasing patient age was observed for higher risk tertiles. However, unlike many other studies, in our analysis, age did not emerge as an independent risk predictor. This was also shown in a study of untreated post-MI patients (24), in which patient age was associated with in-hospital mortality but did not predict death after discharge. The adjustment for estimated creatinine clearance, which includes age, may have influenced this finding.
Risk profile of the OAT cohort
One-year mortality in OAT was 4.3%, and the composite of death, heart failure, or infarction occurred in 9.2% of cases. These rates appear to be substantially higher than 1-year event rates reported previously for stable angina patients in the Euro Heart Survey (mortality 1.5%, composite event rate 3.8%) and more similar to 6-month mortality in ACS patients in the GRACE (Global Registry of Acute Coronary Events) registry (4.8%) or in acute ST-segment elevation MI patients treated invasively (4.0%) (14,25). Thus, despite being relatively stable for a post-MI cohort, OAT patients overall were at characteristically high risk, comparable to that of acute ACS treated in a timely manner. In addition, the shape of the event curves, marked by a steep rise in the early period and stabilization at a lower, constant rate thereafter, are more comparable to ACS than to stable angina. The finding of a strong relationship between the number of days since the qualifying MI to study entry and the risk of events in the current study highlights the timing of highest risk very early after MI.
Lack of PCI benefit across the risk strata: clinical implications
The subset (over 700 patients) with the highest risk of cardiovascular events had a high prevalence of the left anterior descending artery as the culprit artery, diabetes, signs of heart failure during or after the index MI, or relatively low LVEF. The lack of interaction between risk tertile and treatment effect suggests that none of the large number of characteristics studied can identify patients who would benefit from PCI of the persistently occluded IRA in stable patients with 1- to 2-vessel disease. As noted in the original OAT report, predefined subgroup variables of interest, including time from index MI, culprit vessel, baseline EF, diabetes, heart failure, and severity of coronary disease did not interact individually with treatment (3). These findings also suggest that enrollment of more patients with low EF or heart failure, or closer to 24 h after MI would not have altered the findings of the OAT study. Based on the current analysis, routine interventional treatment of the occluded IRA in stable patients without significant spontaneous or inducible ischemia days to weeks after MI should be discouraged.
The main limitations of our study are those inherent to the subgroup analyses, which included multiplicity, smaller size, and confounding. To compensate for these limitations, we applied statistical models with careful control for type I error rate, which should have limited the chance of finding false predictors. However, due to application of our statistical methods, we may have missed some predictors that are clinically significant.
Importantly, these results are only applicable to patients who would have qualified for inclusion in the OAT study. The results do not apply to post-MI patients with severe heart failure, electrical instability, angina at rest, severe inducible ischemia, or 3-vessel disease.
The occurrence of adverse events in patients with persistent total IRA occlusion after MI who are clinically stable is associated with easily identifiable clinical variables, which may guide further medical management. Late recanalization of the IRA does not reduce the rate of clinical events after MI regardless of baseline risk and may be associated with increased risk in the highest risk patients.
The authors thank the patients who enrolled in the study, their physicians, and the staff at the study sites for their important contributions; the staff at the coordinating centers and core laboratories for their hard work; and Erika Laurion for assistance in the preparation of the manuscript.
This study was supported by National Heart, Lung, and Blood Institute grants to Drs. Hochman (U01 HL062509) and Barton (U01 HL062511). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Heart, Lung, and Blood Institute or the National Institutes of Health.
- Abbreviations and Acronyms
- acute coronary syndrome
- ejection fraction
- infarct-related artery
- left ventricular
- left ventricular ejection fraction
- myocardial infarction
- New York Heart Association
- percutaneous coronary intervention
- Received July 9, 2008.
- Accepted August 5, 2008.
- American College of Cardiology Foundation
- Antman E.M.,
- Anbe D.T.,
- Armstrong P.W.,
- et al.,
- ACC/AHA Task Force on Practice Guidelines
- Van de Werf F.,
- Ardissino D.,
- Betriu A.,
- et al.,
- Task Force on the Management of Acute Myocardial Infarction of the European Society of Cardiology
- Widimsky P.,
- Bilkova D.,
- Penicka M.,
- et al.
- Dzavik V.,
- Buller C.E.,
- Lamas G.A.,
- et al.
- Boersma E.,
- Pieper K.S.,
- Steyerberg E.W.,
- et al.,
- PURSUIT Investigators
- Clayton T.C.,
- Lubsen J.,
- Pocock S.J.,
- et al.
- Daly C.A.,
- De Stavola B.,
- Sendon J.L.,
- et al.
- Morrow D.A.,
- Antman E.M.,
- Charlesworth A.,
- et al.
- Schiffrin E.L.,
- Lipman M.L.,
- Mann J.F.
- Stein B.,
- Weintraub W.S.,
- Gebhart S.P.,
- et al.
- Broderick J.P.,
- Phillips S.J.,
- O'Fallon W.M.,
- Frye R.L.,
- Whisnant J.P.
- Golomb B.A.,
- Dang T.T.,
- Criqui M.H.
- Kallikazaros I.,
- Tsioufis C.,
- Sideris S.,
- Stefanadis C.,
- Toutouzas P.