Aims The Occluded Artery Trial (OAT) (n = 2201) showed no benefit for routine percutaneous intervention (PCI) (n = 1101) over medical therapy (MED) (n = 1100) on the combined endpoint of death, myocardial infarction (MI), and class IV heart failure (congestive heart failure) in stable post-MI patients with late occluded infarct-related arteries (IRAs). We evaluated the potential for selective benefit with PCI over MED for patients enrolled early in OAT.
Methods and results We explored outcomes with PCI over MED in patients randomized to the ≤3 calendar days and ≤7 calendar days post-MI time windows. Earlier, times to randomization in OAT were associated with higher rates of the combined endpoint (adjusted HR 1.04/day: 99% CI 1.01–1.06; P < 0.001). The 48-month event rates for ≤3 days, ≤7 days post-MI enrolled patients were similar for PCI vs. MED for the combined and individual endpoints. There was no interaction between time to randomization defined as a continuous (P = 0.55) or categorical variable with a cut-point of 3 days (P = 0.98) or 7 days (P = 0.64) post-MI and treatment effect.
Conclusion Consistent with overall OAT findings, patients enrolled in the ≤3 day and ≤7 day post-MI time windows derived no benefit with PCI over MED with no interaction between time to randomization and treatment effect. Our findings do not support routine PCI of the occluded IRA in trial-eligible patients even in the earliest 24–72 h time window.
The Occluded Artery Trial (OAT) failed to show a clinical benefit with routine mechanical recanalization of the occluded infarct-related artery (IRA) in stable patients enrolled 3–28 calendar days following an index myocardial infarction (MI) on the clinical endpoints of death, MI, and class IV congestive heart failure (CHF).1 This lack of benefit was not related to peri-procedural risk from percutaneous intervention (PCI) and was contrary to prior experimental and non-randomized clinical observations.2–9 There was, however, a broad time window for entry into OAT that spanned as early as just over 24 h post-MI. This earliest time window is close to the latest times for which clinical benefit has been reported with primary angioplasty in observational registries.10,11 In a randomized clinical trial, a routine invasive strategy (largely PCI with concomitant utilization of abciximab) has been reported to result in modest reductions in infarct size in asymptomatic individuals 12–48 h after MI.12 Thus, the overall negative results of the OAT trial warrant further analysis. Late salvage was a proposed mechanism of benefit in OAT that could have been obscured by the broad window of timing of enrolment, with many patients being randomized late in the eligibility time window. The median time from index MI to study randomization in this trial was 8 days, and the 28 day trial eligibility window could thus have resulted in a heterogeneous enrolled population. The potential for significant myocardial salvage and early attenuation of LV remodelling led us to explore the clinical benefits of PCI in patients enrolled in the early (≤3 calendar days post-MI) and intermediate enrolment time windows (≤7 days) in the trial.
The design, methodology, and results of OAT have been previously reported.1,13 In brief, the study randomized 2201 stable subjects with an occluded IRA, 3–28 days following an index MI either to a test strategy of PCI of the IRA with optimal medical therapy (PCI) or to a control strategy of optimal MED alone.1 Thirty-five patients were enrolled in 2006 to complete a viability ancillary study and were added to the 2166 previously reported patients enrolled through 2005. All participants were stable at trial enrolment with a qualifying angiogram performed within the 3–28 day post-MI window showing TIMI 0/1 flow in the culprit IRA. Subjects with cardiogenic shock, NYHA class III/IV heart failure, rest angina or severe ischaemia on stress testing, significant left main or three-vessel epicardial coronary artery disease were ineligible for trial participation. Patients were required to have an ejection fraction (EF) <50% and/or a culprit occlusion in a proximal epicardial vessel to be considered ‘high risk’ and trial eligible. Subjects were followed for a mean duration of 3.2 years, with the primary endpoint being a composite of all-cause death, recurrent MI, and development of NYHA class IV heart failure that was centrally adjudicated. All sites obtained local IRB approval for study conduct.
The eligibility window of 3–28 days post-MI was protocol-defined and based on the number of calendar days following the date of index infarction (day 1). As a result, the earliest time to randomization after MI symptom onset could be as early as 24 h (23.59 p.m., day 1–00.01 a.m., day 3) post-MI. This calendar day criteria for the timing of enrolment was utilized instead of a 24 h post-MI minimum time to avoid the enrolment of a large proportion of OAT patients within the 48 h period. The rate of spontaneous reperfusion is highest in that time period and would potentially dilute a PCI treatment effect. A recent study confirmed that almost half of the untreated patients with ST elevation MI have spontaneous opening of the IRA between 12 and 48 h post-MI.12 Protocol-assigned PCI could occur immediately following randomization but was mandated within 24 h.
We compared baseline characteristics for enrolled patients assigned to PCI and MED alone both in the early (≤3 days) and intermediate (≤7 days) time windows following the onset of qualifying MI. The latter time period was protocol-pre-specified and the former time subset was specified prior to these analyses. We utilized a χ2 test for discrete variables and a Student’s t-test for continuous variables. Outcomes of death, MI, or occurrence of class IV CHF are reported as estimates of 48 month event rates. These estimates were calculated using the Kaplan–Meier product-limit method. Patients lost to follow-up were censored at the time of last available contact. We performed an analysis of the primary outcome for subjects enrolled both at the early (≤3, >3 days) or intermediate time (≤7, >7 days) points utilizing Cox proportional regression. To adjust for multiple comparisons in OAT, all non-primary analyses are considered significant only at an alpha level of 0.01, including tests for interaction. The covariate-adjusted hazard ratio was quantified utilizing a Cox proportional hazards model regression model utilizing a risk model that included demographic, clinical, and angiographic variables.14
Of the 2201 patients enrolled, 331 were randomized early (≤3 days post-MI); 168 were assigned to PCI and 163 to MED alone. Exact time in hours from index MI onset was available in 293/331 early subjects. Among this group, 38 patients were enrolled <24 h, 150 patients within 24–48 h, and 105 within 48–72 h of their presenting infarction. The mean time from index MI to randomization was 2.6 ± 0.5 calendar days for the 168 patients assigned PCI (inter-quartile range 2–3 days) and 2.7 ± 0.5 days (inter-quartile range 2–3 days) for the MED group.
Similarly, 975 subjects were enrolled ≤7 days following MI, with 492 assigned to PCI and 483 to MED alone. Baseline characteristics for patients enrolled in the early and intermediate time windows were compared and then stratified by assigned treatment (PCI vs. MED) (Table 1). Subjects enrolled ≤3 calendar days post-MI were quite similar to patients enrolled >3 days with few exceptions. Patients randomized in the early time window were more likely to be current smokers (46 vs. 38%, P = 0.005) and to have a family history of coronary artery disease (49 vs. 39%, P = 0.0004). Resting heart rate at enrolment (74 ± 12 vs. 71 ± 12 b.p.m., P < 0.0001) and baseline fasting glucose (130 ± 46 vs. 118 ± 41 mg/dL, P < 0.0001) were also significantly higher in this group. Similar statistically significant but clinically small differences in baseline characteristics are also seen between the subsets of patients enrolled ≤7 and >7 days. Of note, patients enrolled >7 days after their index MI were more likely to have collaterals to the IRA documented on their qualifying angiogram (86 vs.91%, P = 0.0007).
Patients assigned to PCI were similar to those randomized to medical treatment at all time windows except for higher rates of diabetes in the MED group (>3 day and ≤7 day groups), as seen in the overall trial. The location and status of the IRA, presence of collaterals to the IRA, and time from MI to randomization were similar at all time intervals studied.
Percutaneous intervention outcomes
Protocol-mandated PCI was performed on 166/168 patients assigned to the early time window (≤3 days post-MI) at a median of 3.0 calendar days after index MI (inter-quartile range 2–3 days). Procedural success rates were not influenced by the time from index MI to randomization (Table 2). Post-procedural TIMI flow rates were similar for all the groups analysed. The overall risk of peri-procedural MI was small despite protocol-mandated post-procedural CPK/CKMB sampling in all patients. No trends to suggest a time-dependence of procedural risk are noted.
Assigned percutaneous intervention outcomes for patients enrolled in the 3 and 7 day time windows in the Occluded Artery Trial percutaneous intervention-assigned group
≤3 days (n = 165)
>3 days (n = 913)
≤7 days (n = 478)
>7 days (n = 600)
Pre-PCI TIMI grade flow (%)
Post-PCI flow (%)
Days from MI–PCI
3 ± 0.88
13 ± 7.6
4.8 ± 1.9
17.0 ± 6.7
PCI-related MI (n)
LAD, left anterior descending; RCA, right coronary artery.
Discharge medication and follow-up medication at 1 year for early-time-window patients in the PCI and MED arms are illustrated in Table 3. Overall medical treatment was consistent with guideline recommendations in both PCI- and MED-assigned patients. The use of a thienopyridine was significantly higher in PCI-assigned subjects. There is a statistically significant but numerically small increase in aspirin utilization in the PCI-assigned group. Similar trends in medical management (not shown) were also noted for patients randomized in the intermediate time window.
Medications at discharge and 1 year for patients randomized to percutaneous intervention and optimal medical therapy ≤3 days and >3 days following index myocardial infarction
≤3 days (n = 331)
>3 days (n = 1870)
n = 168 (%)
n = 163 (%)
n = 933 (%)
n = 937 (%)
ACE or ARB
n = 157 (%)
n = 141 (%)
n = 834 (%)
n = 847 (%)
At 1 year
ACE or ARB
Earlier, times from MI to randomization were associated with a higher event rate for the primary endpoint (Figure 1). However, there was no interaction between treatment effect and time to randomization treated as a continuous variable (P = 0.55) or as a categorical variable (≤3 days vs. after 3 days, P = 0.77; ≤7 days vs. after 7 days, P = 0.65). In a multivariable model for the primary outcome, the variable time from MI to randomization was an independent predictor (adjusted HR 1.04/day: 99% CI 1.01–1.06; P < 0.001).
Relationship between days from index myocardial infarction to enrolment vs. rate of 4 year composite outcome for all patients enrolled in the Occluded Artery Trial. Patients enrolled the earlier allowable time window were more likely to experience the primary outcome during follow-up (P < 0.001). There was no interaction between time from myocardial infarction to randomization (days) and treatment effect (P = 0.55).
Estimated 48 month event rates and Kaplan–Meier curves comparing the composite endpoint as well as its individual components are depicted in Table 4 and Figure 2, respectively. There was no benefit with PCI over MED on the combined endpoint of death, non-fatal MI, and class IV heart failure for patients enrolled ≤3 days following index MI. Individual components of the combined endpoint are also similar between PCI and MED. Similarly, there was no benefit with PCI over MED on the combined endpoint of death, non-fatal MI, and class IV for the larger group of patients enrolled ≤7 days following index MI.
Kaplan–Meier analysis of cumulative incidence of the composite endpoint of death, myocardial infarction, and class IV heart failure as well as the individual components for percutaneous intervention (solid line) and medical therapy groups (broken line) for patients enrolled in the ≤3 and ≤7 day subgroups.
Forty-eight month primary and secondary outcomes by treatment assignment in the Occluded Artery Trial for patients randomized in the early and intermediate time points following myocardial infarction
PCI group [≤3 days (n = 168); >3 days (n = 933)]
MED group [≤3 days (n = 163); >3 days (n = 937)]
Treatment interaction (3 days)
PCI group [≤7 days (n = 492); >7 days (n = 609)]
MED group [≤7 days (n = 483); >7 days (n = 617)]
Treatment interaction (7 days)
Treatment interaction (QMI days)
Class IV CHF
Site non-fatal MI
QMI days, days from MI to randomization.
Consistent with the overall trial findings, PCI did not reduce clinical events in OAT patients enrolled in the earliest time window of trial eligibility. Enrolment in the earliest time period post-MI was independently associated with higher event rates, but we observed no interaction between time to randomization and treatment effect. The lack of benefit with PCI was also noted when time was analysed as a categorical variable using an early (≤3 days) and intermediate (≤7 days) cut-off. This analysis supports the clinical application of the results of the parent trial for the management of stable patients with documented total occlusion over the entire study enrolment period of 24 h to 28 days.
This analysis has important implications. Late presentation of stable patients following STEMI, beyond the guideline recommended 12 h window following symptom onset, is not uncommon.15,16 There is a paucity of randomized clinical trial data on mechanical recanalization in the 24 h–7 days window post-MI, and the current analysis which demonstrates a lack of clinical benefit in patients randomized in this time period in OAT contributes to the evidence. Our observations suggest that there is unlikely to be any significant benefit for a strategy of routine PCI performed in asymptomatic, haemodynamically stable late presenters with an occluded IRA >24 h post-MI onset when compared with a strategy of optimal MED alone. It is important for the clinician to apply these findings to patients meeting OAT eligibility and not to those with ongoing chest pain, rest or low-threshold angina, severe ischaemia, haemodynamic instability (cardiogenic shock or NYHA class III/IV), left main disease, or three-vessel disease.
The smaller infarct size observed at 5–10 days with a routine invasive strategy 12–48 h post-MI in the mechanistic BRAVE 2 study may appear to be at odds with the lack of clinical benefit with PCI in the early OAT subset. However, the correlation between infarct size assessed at this time point and clinical events over 3 years in a population with persistent total occlusion of the IRA often with angiographically visible collaterals is unknown. Furthermore, there were fundamental differences in the patients enrolled in these trials. By design, patients in OAT were required to have documented total occlusion of the IRA (TIMI 0/1 flow), stress testing was required when significant residual ischaemia was suspected (any retained wall motion in the IRA distribution), and patients with severe provocable ischaemia were excluded. In contrast, 43% of subjects in the invasive arm of BRAVE 2 had an open artery with TIMI 2/3 flow in the IRA, prior to intervention; stress testing was not performed before randomization, and consequently, patients with severe inducible ischaemia were not excluded, perhaps explaining the 33% rate of unplanned PCI during the initial 30 day follow-up period in contrast to 6.6% (73/1100) in the OAT MED arm. In fact, in the BRAVE 2 population with myocardium in jeopardy, the only multivariate predictor of final smaller infarct size in the invasive group was superior initial TIMI flow grade, even when final TIMI flow grade was entered into the model. This finding appears to suggest that the reduction of infarct size in BRAVE 2 was possibly due to the prevention of re-occlusion instead of recanalization salvaging myocardium in the invasive strategy group. Finally, BRAVE 2 had limited follow-up and was not powered to assess clinical outcomes. The early post-MI OAT cohort we report here is substantially larger than the BRAVE 2 subset with total occlusions and no clinical benefit in the setting of this rigorous randomized clinical trial with over 3 years mean clinical follow-up. Of note, OAT patients with mild–moderate ischaemia in the territory of the IRA also derived no benefit from protocol-assigned PCI.17
The lack of benefit noted with PCI in the OAT cohort enrolled within 3 days post-MI is intriguing. In contrast to the classical wavefront paradigm of Reimer and Jennings in the animal model,18 the presence of collateral circulation and the protective effect of ischaemic pre-conditioning lead to the presence of viability in the infarct and peri-infarct distribution in the extended post-MI period.19,20 The nuclear ancillary OAT also confirms at least moderate retained viability in the region of the index MI in 69% of patients, and the significant improvement in EF in those with viability was similar in PCI and MED groups.21 Those findings and the current report suggest that the presence of viability alone does not translate into meaningful clinical benefit with PCI in this clinical situation even in analyses restricted to the early and intermediate OAT post-MI time windows. The lack of benefit observed is also not explained by procedural risk, as both death and procedural MI in the PCI population were negligible. Finally, the lack of benefit for PCI cannot be attributed to demographic, clinical, and treatment characteristics, as these were similar and adjusted for in this analysis. It remains possible that clinical benefits from PCI, though present, are inconsequential in magnitude and do not favourably impact on the overall prognosis of the index MI in this patient population. The 95% confidence intervals in OAT include a possible 8% benefit with PCI but also 43% harm.
The median time from MI randomization in the OAT trial was 8 (5, 16) days. The intermediate time point ≤7 days was pre-specified in the original protocol and represents 44% (975/2201) of the study population. Although the lack of treatment effect with PCI in this group was briefly mentioned in the original report, details of this subgroup are presented in this manuscript. The independent association of early time to randomization and excess clinical events confirms that the highest risk period is early post-MI. However, there is no suggestion of benefit with PCI for this subset, which is similar to the findings for the highest risk OAT tertile.14
OAT was adequately powered to detect a benefit for the tested PCI strategy. Our subset analysis is however underpowered to definitively confirm a lack of treatment effect in this early and intermediate time period post-MI. There is however no suggestion of favourable outcome with PCI noted utilizing time to enrolment both as a continuous and as a dichotomous variable. In the context of a rigorously performed clinical trial, these subsets of patients in the early and intermediate time points represent the largest available data set to date. Extended long-term follow-up is currently underway to examine potential late trends. Our findings are consistent with the observation that the risk period for events is highest soon after coronary occlusion. But in contrast to our findings, routine PCI up to 12 h post-STEMI reduces death. By design, OAT was designed to exclude the early unstable period, which is also the period with the highest rate of spontaneous coronary artery recanalization. Patients evaluated for OAT >24 h post-MI who had an event or were unstable before angiography were not enrolled and the results do not apply to this patient subset.
We observed no benefits with routine PCI of the IRA in OAT patients enrolled ≤3 and ≤7 days post-MI. This lack of benefit supports optimal MED alone as the initial management strategy in this patient population. Late routine mechanical recanalization of the IRA cannot be supported in this population.
National Heart, Lung, and Blood Institute, National Institutes of Health (U01 HL062509, U01 HL062511).
; for the Survival Ventricular Enlargement Investigators. Effect of infarct artery patency on prognosis after acute myocardial infarction. The Survival and Ventricular Enlargement Investigators. Circulation 1995;92:1101-1109.
. Primary angioplasty versus no reperfusion therapy in patients with acute myocardial infarction and a pre-hospital delay of >12–24 hours: results from the pooled data of the maximal individual therapy in acute myocardial infarction (MITRA) registry and the myocardial infarction registry (MIR). J Invasive Cardiol 2001;13:367-372.
. Predictors of outcome and the lack of effect of PCI across the risk strata in patients with persistent total occlusion after myocardial infarction: results from the Occluded Artery Trial (OAT). Circulation 2007;116:625-625.
. Practice variation and missed opportunities for reperfusion in ST-segment-elevation myocardial infarction: findings from the Global Registry of Acute Coronary Events (GRACE). Lancet 2002;359:373-377.
. Prospective evaluation of clinical outcomes after acute ST-elevation myocardial infarction in patients who are ineligible for reperfusion therapy: preliminary results from the TETAMI registry and randomized trial. Circulation 2003;108Suppl. 1:III14-III21.
. Impact of stress testing prior to PCI or medical management on outcomes of patients with persistent total occlusion after myocardial infarction: analysis from the Occluded Artery Trial (OAT). J Am Coll Cardiol 2008;51:A191.
. The Occluded Artery Trial (OAT) viability ancillary study (OAT-NUC): influence of infarct zone viability on left ventricular remodeling after PCI vs. medical therapy alone. Circulation 2007;116:624-625.