Lack of ST-segment resolution despite attainment of TIMI (Thrombolysis In Myocardial Infarction) flow grade 3 in the epicardial coronary artery is associated with an increased mortality risk, and ST-segment resolution is considered a marker for microvascular perfusion (1). Prior studies have shown that TIMI flow grade 3 is obtained in more patients after primary percutaneous coronary intervention (pPCI) than after fibrinolysis (1- 2); however, the relationship between the mode of reperfusion and ST-segment resolution is uncertain. Using data from the largest randomized trial comparing pPCI versus fibrinolysis, we hypothesized that early ST-segment resolution is more complete with pPCI, and the prognostic value of ST-segment resolution differs with the mode of revascularization being mechanical or pharmacological.

The DANAMI-2 (DANish trial in Acute Myocardial Infarction-2) randomly assigned 1,572 patients with ST-segment elevation myocardial infarction (STEMI) to either pPCI or fibrinolysis with front-loaded tissue plasminogen activator (2). The present substudy was conducted in patients with an available randomization 12-lead electrocardiogram (ECG) with complete ST-segment measurements. The DANAMI-2 study complied with the Declaration of Helsinki, and the local Ethics Committee and Danish Data Protection Agency approved data collection.

A standard 12-lead ECG was performed in all patients at randomization (baseline ECG). Additional ECGs were obtained immediately before start of reperfusion treatment, and as timely, as clinically feasible at 90 min and 4 h after initiation of reperfusion treatment. The ECGs were analyzed at a centralized core laboratory blinded to all patient data. The ST-segment deviation was measured manually in the J-point to the nearest 0.5 mm in all leads except aVR. The sum of ST-segment elevation (ΣST↑) in all leads was determined for the individual ECG, and ST-segment resolution was assessed by the reduction in ΣST↑ between the baseline ECG and the following ECG as the percentage of baseline ΣST↑. Patients were divided into 3 groups according to the amount of ST-segment resolution: 1) complete ST-segment resolution (≥70%); 2) partial ST-segment resolution (30% to <70%); and 3) no resolution (<30%) (3).

The end points were 30-day and long-term mortality from any cause or clinical reinfarction (2). Median follow-up was 4.2 years for mortality and 3.0 years for reinfarction. No patients were lost to follow-up.

Baseline dichotomous variables were compared using the chi-square test and presented as percentages, and continuous variables were compared using a rank-sum test and presented as median with interquartile range (IQR). The rate of events in the 3 different ST-segment resolution categories at 90 min and 4 h were analyzed using the trend test. Kaplan-Meier curves were compared using log-rank statistics. Univariate analysis was performed for all baseline variables, and significant variables were included in a multivariable backward selection model excluding all variables with a p value >0.05. The Cox proportional hazard model was used to determine predictors of outcome. The ST-resolution was entered into the Cox model as a continuous variable and hazard ratio given per ST-segment resolution increase of 1%. Interactions were assessed by entering the interaction term into the Cox model. A p value (2-sided) ≤0.05 was considered significant. The SAS software, version 8.0 (SAS Institute, Cary, North Carolina) was used.

Demographics for the 1,421 included patients with a baseline ECG are presented in (Table 1). The ECGs recorded at pre-intervention, at 90 min, and at 4 h were available for 880, 1,233, and 1,208 patients, respectively.

(Figure 1) demonstrates median percent ST-segment resolution over time in the 2 treatment groups. Before the reperfusion attempt, more ST-segment resolution occurred in patients randomly assigned to pPCI (12%, IQR 0% to 31%, vs. 6%, IQR 0% to 47%; p = 0.004). Pronounced normalization of the ST-segment was observed in both groups at 90 min, but significantly more after pPCI (60%, IQR 33% to 82%, vs. 45%, IQR 13% to 71%; p < 0.0001). This difference had faded at 4 h (65%, IQR 40% to 80%, vs. 63%, IQR 35% to 85%; p = 0.2).

Patient demographics according to ST-segment resolution at 4 h are listed in (Table 2). Patients without ST-segment resolution tended to have the highest mortality rate regardless of treatment strategy; however, this difference only reached statistical significance in the fibrinolysis group ((Table 3),Figure 2). By multivariable analysis, age, diabetes mellitus, congestive heart failure, and ST-segment resolution at 4 h were found to be strong predictors of long-term mortality in all patients (Table 4). A difference in 1% ST-segment resolution increased the mortality risk by 8%. When dividing patients according to treatment group, 4-h ST-segment resolution and time-to-treatment were only independent predictors of mortality in the fibrinolysis group. However, no interaction was found between ST-segment resolution and treatment strategy regarding mortality (p = 0.4).

In the fibrinolysis group, reinfarction rates were significantly higher among patients with complete ST-segment resolution, but there was no association between reinfarction and ST-segment resolution for patients receiving pPCI (Table 5). Multivariable analysis of all patients showed that age, diabetes, heart failure, and time to treatment were independent predictors of reinfarction (Table 4). In the fibrinolysis group, the risk of reinfarction increased 1% with each additional 1% of ST-segment resolution, whereas in the pPCI group, ST-segment resolution did not influence reinfarction rate. There was a significant interaction between ST-segment resolution and treatment strategy regarding reinfarction (p = 0.01).

This study showed that median percent ST-segment resolution at 90 min after reperfusion therapy is more complete with pPCI than with fibrinolysis. However, a catch-up phenomenon occurred at 4 h, resulting in similar levels of ST-segment resolution. Interestingly, ST-segment resolution was only predictive of mortality after fibrinolysis and not after pPCI. Additionally, complete ST-segment resolution was predictive of increased reinfarction among patients receiving fibrinolysis.

The median interval between randomization and initiation of reperfusion therapy was 49 min longer with pPCI than with fibrinolysis, inherent to the transportation delay (2). However, time from randomization to artery patency was still similar in the 2 treatment arms because needle to reperfusion time is approximately 45 min with fibrinolysis (4) and momentarily with pPCI. During the time delay, pPCI patients obtained more ST-segment resolution, which could be explained by either spontaneous regression or likely induced by aspirin and heparin intake at the referral hospital before transfer. On the contrary, patients randomly allocated to fibrinolytic therapy were treated with aspirin and heparin immediately before administration of fibrinolysis. This timing difference could favor the increased ST-segment resolution seen in the pPCI group at 90 min.

Our results are supported by an earlier study reporting a significant difference in 2-h ST-segment resolution with pPCI versus fibrinolysis (5). The finding of more patients who were male, and had diabetes, prior myocardial infarction, and anterior myocardial infarction in the group of patients without ST-segment resolution is consistent with earlier findings (3), in which patients without ST-segment resolution tended to have the worst risk profile at baseline. It is noteworthy that these high-risk patients had the least amount of ST-segment elevation at baseline. That may be caused by a difference in etiology, with these patients tending to have general arthrosclerosis with collaterals and a smaller culprit vessel than patients with complete ST-segment resolution, who may have a large epicardial artery with acute occlusion.

Dividing ST-segment resolution into 3 groups is a frequently used method for determining the relationship between treatment and mortality (3,6). Corresponding with our results, van't Hof et al. (6) found that 51% had complete ST-segment resolution 1 h after pPCI; however, we could not show a significant association with mortality. One explanation might be the use of stents in 93% of patients in the present study versus none in the study by van't Hof et al. (6). The placement of stents with high pressure can induce distal embolization, which could influence ST-segment resolution, but not mortality.

Traditionally, ST-segment resolution is determined 90 to 180 min after the start of fibrinolysis (1,7), whereas the optimal timing for determining ST-segment resolution preceding pPCI is uncertain (8) and may be at a much earlier time, given the immediate artery patency with pPCI. That may explain why we, in this study, with the earliest available post-procedure ECG recorded after 90 min, did not detect a relationship between ST-segment resolution and outcome in the pPCI group, while both 90-min and 4-h ECG predicted outcome after fibrinolysis. Consequently, both 90-min and 4-h ECG remain an important predictor in fibrinolysis, whereas early continuous ECG monitoring after pPCI may be necessary to rely on the findings of future trials (8).

Our finding of complete ST-segment resolution being associated with an increased rate of reinfarction in patients treated with fibrinolysis is new. An explanation may be that patients with complete ST-segment resolution are experiencing a large acute occlusion. When the occlusion is removed by fibrinolysis, it leaves a vulnerable plaque prone to reinfarction later. Conversely, in patients without ST-segment resolution, the microcirculation is disrupted, leaving no or only minor viable myocardium exposed for later reinfarction. The consequence of reinfarction after the index infarct is a higher mortality rate (9), making any method that can identify patients at increased risk of reinfarction important. Our results suggest that all patients receiving fibrinolysis may benefit from catheterization. Accordingly, rescue PCI in patients with failed fibrinolysis has been shown not only to reduce reinfarction rates but also to lower heart failure and mortality (10). Thus, it is tempting to speculate whether fibrinolysis followed by PCI could be an alternative to pPCI in areas where pPCI is not ready available. However, caution should be taken regarding both timing and which fibrinolytic regime to use, as facilitated PCI has been shown to increase stroke rate, ischemic cardiac complications, heart failure, shock, and mortality rates (11). On the contrary, the CARESS (Combined Abciximab REteplase Stent Study) trial (12) found improved outcome for STEMI patients immediately transferred for PCI after half-dose reteplase plus abciximab, compared to management at the local hospital.

ST-segment resolution was measured at different time points after the presumed time of an open artery in the 2 treatments groups. This difference could have biased the findings in favor of pPCI. An ECG immediately after PCI may have been preferable and more comparable to a 90-min ECG in patients receiving fibrinolytic agents. Additionally, continuous ST-segment monitoring would have been preferable as opposed to the static standard 12-lead ECG because within-patient variability may confound the present observations.

Our results indicate that ST-segment resolution in the conventional 90-min and 4-h ECG remains an important prognostic factor after fibrinolysis, and confirm the clinical utility of these ECGs in selecting patients for rescue PCI. The provocative finding of limited prognostic value of ST-segment resolution in a population of STEMI patients treated with pPCI deserves further investigation, especially because it is already widespread as a surrogate end point in trials.