Author + information
- Received April 22, 2004
- Revision received June 9, 2004
- Accepted June 14, 2004
- Published online September 15, 2004.
- Michael G. McLaughlin, MD*,
- Gregg W. Stone, MD, FACC†,
- Eve Aymong, MD, MS†,
- Graham Gardner, MD*,
- Roxana Mehran, MD, FACC†,
- Alexandra J. Lansky, MD, FACC†,
- Cindy L. Grines, MD, FACC‡,
- James E. Tcheng, MD, FACC§,
- David A. Cox, MD, FACC∥,
- Thomas Stuckey, MD, FACC¶,
- Eulogio Garcia, MD#,
- Giulio Guagliumi, MD**,
- Mark Turco, MD, FACC††,
- Mark E. Josephson, MD, FACC* and
- Peter Zimetbaum, MD, FACC*,* ()
- ↵*Reprint requests and correspondence:
Dr. Peter Zimetbaum, Beth Israel Deaconess Medical Center, One Deaconess Road, Baker 4, Boston, Massachusetts 02215
Objective This study was done to assess and compare the prognostic significance of multiple methods for measuringST-segment elevation resolution (STR) following primary percutaneous coronary intervention (PCI).
Background Resolution of ST-segment elevation (STE) is a powerful predictor of both infarct-related artery patency and mortality in acute myocardial infarction (AMI). Recent thrombolytic studies have suggested that simple measures of STR may be as powerful as more complex algorithms. The optimal method of assessing STR following primary PCI has not been studied.
Methods We analyzed 700 patients with technically adequate baseline and post-PCI electrocardiograms from the Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) trial. Five methods were used to assess STR: 1) summed %STR across multiple leads (ΣSTR); 2) %STR in the single lead with maximum baseline STE (MaxSTR); 3) absolute maximum STE before the procedure; 4) absolute maximum STE after intervention (MaxSTPost); and 5) a categorical variable based upon MaxSTPost (High Risk).
Results At 30 days, ΣSTR, MaxSTR, and MaxSTPost all correlated strongly with mortality (p = 0.004, p = 0.005, and p < 0.0001, respectively) and the combined end point of mortality or reinfarction (p = 0.001, p = 0.001, and p < 0.0001). At one year, ΣSTR and MaxSTPost correlated with mortality (p = 0.04, p = 0.0001), reinfarction (p = 0.02, p = 0.0015), and the combined end point (p = 0.02, p < 0.0001). By multivariate analysis, only the simpler measures of MaxSTPost and High Risk categorization independently predicted all outcomes at both time points.
Conclusions The STR following primary PCI in AMI correlates strongly with mortality and reinfarction, independent of target vessel patency. The simple measure of the maximal residual degree of STE after primary PCI is a strong independent predictor of both survival and freedom from reinfarction at 30 days and 1 year.
Assessment of ST-segment elevation resolution (STR) after thrombolytic therapy has established utility for predicting both target vessel patency and prognosis in acute myocardial infarction (AMI) (1–14). A number of methods for measuring STR have been studied, mostly involving percent resolution of ST-segment deviation. Some use a single electrocardiogram (ECG) lead (2,9,11,15), whereas others use sum deviation across multiple leads (6–8,10,12,13,16,17); some include ST-segment elevation (STE) alone (2,7,8,17) and others include reciprocal depression (6,12,13). Few studies have directly compared the various methods (11,18,19). The patterns and significance of STR after primary percutaneous coronary intervention (PCI) have been less extensively studied (9,10,16,20,21) than those after thrombolysis. Furthermore, the relative prognostic value of different measures of STR has not been evaluated after mechanical reperfusion therapy. Since ST-segment changes correlate with epicardial vessel flow rates (2,4,5,8,12), which are more commonly normalized after primary PCI than after thrombolysis, ST-segment evolution may be more homogeneous and, therefore, less predictive after primary PCI. Additionally, the extent and pattern of myocardial reperfusion, which has also been shown to correlate with both STR and outcomes (21,22), may differ after mechanical and pharmacological reperfusion. Therefore, we examined the predictive accuracy and clinical utility of several different scoring methodologies of STR from a large prospective, controlled study of patients with AMI undergoing primary PCI.
Patient population and study protocol
The Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) trial protocol has been previously described (23). In brief, non-shock patients with AMI with symptom duration of <12 h and STE in two or more contiguous leads or left-bundle branch block (LBBB) were consented. Patients with other ECG patterns were also enrolled if angiography showed a high-grade coronary stenosis with an associated left ventricular wall motion abnormality. Following angiography, patients with coronary anatomy suitable for stenting were randomized in a 2×2 factorial design to balloon angioplasty versus stenting, each ±abciximab. A total of 2,082 patients were randomized at 76 international centers.
Inclusion in the formal STR substudy required the following: 1) paired ECGs at baseline and within 4 h after PCI; 2) ≥1 mm STE in two contiguous leads in the infarct territory on the baseline ECG; and 3) absence of conditions on both ECGs that would confound interpretation, including LBBB, pacing, pre-excitation, ectopy, missing leads, artifact, and multiple ECG AMI territories. Anterior myocardial infarction (MI) was defined as occurring when STE was present in leads V1to V6, I, and aVL. Inferior MI was defined when STE was confined to leads II, III, aVF, and V5to V6.
All ECG assessments were made at an independent ECG core laboratory at the Beth Israel Deaconess Medical Center. Measurements were made by two readers blinded to clinical and angiographic results. The absolute level of STE was measured with digital calipers to the nearest 0.01 mV, 20 ms after the end of the QRS interval, using the TP-segment as the isoelectric baseline. On the basis of previously validated algorithms (6,7), summed STE was calculated as follows: for anterior MI, the sum of STE in V1to V6, I; and aVL; for inferior MI, the sum of STE in leads II, III, aVF, V5, and V6.
The following five electrocardiographic scoring systems were studied: 1) summed STR (ΣSTR), the percent reduction in the summed score between the pre- and post-procedure ECG; 2) maximum STR (MaxSTR), the percent reduction in the absolute STE in the single lead in the infarct territory with maximum STE on the baseline ECG; 3) MaxSTPre, the absolute maximum STE in a single lead in the infarct territory on the baseline ECG; 4) MaxSTPost, the absolute maximum STE in a single lead in the infarct territory on the post-procedure ECG; and 5) High Risk (for multivariate analysis only), for inferior MI, MaxSTPost >1 mm, and for anterior MI, MaxSTPost >2 mm. For statistical analysis, summed STR parameters were categorized as none (<30%), partial (30% to 70%), and complete (>70%) (6,7). MaxSTPre was categorized as <2, 2 to 3, 3 to 4, and <4 mm, and MaxSTPost was categorized as <1, 1 to 2, and >2 mm.
Independent core angiographic laboratory analysis was performed as previously described by technicians who had no knowledge of clinical outcomes (23).
Categorical variables were summarized as percentages and compared with the Fisher exact testfor two groups and chi-squared tests for three or more groups. Continuous variables are presented as medians with interquartile ranges and were compared using the Kruskal-Wallis nonparametric test or the Wilcoxon two-sample test. All 30-day and 1-year outcomes are summarized as Kaplan-Meier estimates and compared between groups using log-rank tests. The influence of baseline variables, time to reperfusion, and (separately) each of the STR measures on mortality were evaluated with Cox proportional hazards regression, and the results were expressed as hazard ratios with 95% confidence intervals for the STR measures. The other variables included in the multivariate analysis were age, female gender, diabetes, current smoker, hypercholesterolemia, prior MI, prior PCI, prior bypass surgery, Killip class, anterior MI location, three-vessel disease, randomization to stent, randomization to abciximab, pre-procedure Thrombolysis In Myocardial Infarction (TIMI)flow grade 3, post-procedure TIMI flow grade 3, and left anterior descending infarct vessel. To confirm the relative prognostic value of the different ST-segment parameters, we used the -2 log L statistic (minus twice the logarithm of the maximum likelihood) model fit statistic, which represents the summary measure of agreement between the model and the data (similar to the likelihood ratio, but independent of the number of the covariates chosen by the model). Smaller values indicate a more predictive model.
Of 2,082 enrolled patients, 700 met criteria for inclusion in the STR substudy (Fig. 1).The baseline characteristics of the patients included in the STR substudy were similar to those excluded, except that those in the STR substudy were more likely to have an anterior MI and had slightly worse baseline left ventricular function, reduced infarct artery flow at presentation (but not after PCI), and shorter times to hospital arrival and first balloon inflation (Table 1).Patients specifically excluded for technical reasons or artifacts were older (62 vs. 59 years; p = 0.002) and more commonly had left circumflex artery infarction confounding ECG interpretation (18.7% vs. 12.0%; p = 0.01); otherwise, they were similar in all procedural variables and outcomes to the substudy patients. The time from final balloon inflation to the post-procedure ECG was 1.77 ± 0.83 h.
ST-segment parameters: Frequency and relationship with TIMI flow
The baseline incidences of each MaxSTPre category and of each category of the three measures of STR are shown in Figure 2.Of the 700 patients in the study, 146 qualified for High Risk categorization.
A TIMI flow grade 3 was achieved in 96% of all patients. Relationships between the various measures of STR and post-procedure TIMI flow are shown in Figure 3.Only MaxSTPost significantly correlated with post-PCI TIMI flow grade 3.
Impact of ST-segment evolution on death and reinfarction
Mortality in the STR substudy was 2.7% at 30 days and 5.1% at 1 year. These rates are not significantly higher than in the entire CADILLAC population (2.1% and 4.2%, respectively; p = 0.09 and p = 0.17, respectively).
As seen in Table 2,ΣSTR, MaxSTR, and MaxSTPost all correlated strongly with mortality, reinfarction, and the combined end point of mortality or reinfarction at 30 days. At one year, ΣSTR and MaxSTPost again correlated with mortality, reinfarction, and the combined end point. When only the 96% of patients with final TIMI flow grade 3 were analyzed, the results were nearly identical to those for the entire population (Table 3).
MaxSTPre correlated with mortality, reinfarction, and the combined end point at 1 year, but did not correlate with any outcome at 30 days.
Multivariate analyses: Death and reinfarction
As shown in Figure 4,ΣSTR >70%, MaxSTR >70%, and MaxSTPost >1 mm were statistically significant independent predictors of mortality, reinfarction, and the combined end point at 30 days. When MaxSTPre was forced into the multivariate model with MaxSTPost, MaxSTPost remained independently correlated with all outcomes at similar p values and hazard ratios. High Risk categorization also correlated with mortality, reinfarction, and the combined end point.
Only MaxSTPost and High Risk categorization were independently predictive of one-year mortality (Fig. 5).The ΣSTR and MaxSTR inversely correlated with reinfarction and the combined end point, but not with one-year mortality. Of all clinical, angiographic, and ECG parameters included in the multivariate analysis, only MaxSTPost and High Risk categorization correlated with all outcomes at one year. MaxSTPre did not independently correlate with any of the outcomes at either of the time points in the multivariate analysis. As in the 30-day analysis, controlling for pre-procedure STE by forcing MaxSTPre into the multivariate analysis did not change the predictive power of MaxSTPost.
Model fit statistics
The ΣSTR, MaxSTR, MaxSTPost, and High Risk categorization were all statistically significant components of the multivariate models for the occurrence of death, reinfarction, and the combined endpoint, at both 30 days and 1 year. For five of the six outcomes, the model fit statistic for MaxSTPost and High Risk categorization were superior to those for the other measures (Table 4).For the sixth (reinfarction at 30 days), MaxSTPost was similar to the other two and High Risk categorization was superior.
Our analysis, the largest study to date of the significance of STR after mechanical reperfusion therapy in AMI, demonstrates that resolution of STE is a powerful predictor of short- and long-term mortality and reinfarction after primary PCI, despite the achievement of very high rates of TIMI flow grade 3. Measures of absolute as well as relative STR strongly correlated with survival and freedom from reinfarction. Notably, the simplest measures of STR—assessment of absolute STE in a single lead on a single post-intervention ECG—were at least prognostically equivalent to more complex algorithms. Moreover, these simple measures (MaxSTPost and High Risk categorization) were the only STR parameters to correlate with one-year mortality in multivariate analysis, and the only parameters (even including all clinical and angiographic variables) to correlate with all three outcomes (death, reinfarction, and the combined end point) at one year.
Our findings are consistent with and extend the results from recently published thrombolytic therapy trials showing that uncomplicated measures of absolute maximum STE after reperfusion are at least as powerful as more complex algorithms requiring relative percent STR calculations over multiple summed leads (18, 19). The ΣSTR parameter we utilized is based upon one of the most widely studied STR algorithms, which sums STE across leads in the infarct territory and then categorizes percent resolution as complete, partial, or none (6,7). In several large thrombolysis trials, this measure has correlated with TIMI flow, left ventricular function, and mortality. Although powerful, this method is cumbersome in research settings and difficult for the clinician to apply. Schroder et al. (18), the originator of the summed STR algorithm, more recently described a simpler parameter, MaxSTE, that categorizes patients as low- or high-risk based upon maximum ST-segment deviation after thrombolysis. In a large thrombolysis population, MaxSTE predicted short- and medium-term mortality more accurately than the ΣSTR algorithm (19). Although substantially easier to use than %STR algorithms, this method requires stratification by MI location and the extent of pre-procedure STE. More recently, Cooper et al. (19) described a simpler parameter, minimal ST-segment deviation (MSTD), also based upon maximum STE after thrombolysis. This measure predicted target vessel patency, but not mortality, with similar accuracy to ΣSTR. Our High Risk categorization is similar to MSTD categorization, but even less complex. The MSTD requires identifying the lead with maximum STE at baseline and then assessing that lead after thrombolysis. Our High Risk categorization (as well as MaxSTPost) requires only the post-procedure ECG. The ability to derive prognostic utility from a single lead on a single ECG has obvious advantages over serial assessments and complex calculations, in both the clinical and research settings. The High Risk variable thus extends the benefits of the Schroder et al. (18) MaxSTE, the Cooper et al. (19) MSTD, and our MaxSTPost in that it is a simple categorical variable based upon a single lead after treatment, accounting for infarct location to improve predictive accuracy.
To exclude the possibility that post-procedure STE may simply be a reflection of pre-procedure STE (and therefore the magnitude of the MI rather than the efficacy of therapy), we also assessed baseline STE with MaxSTPre. This parameter correlated weakly with outcomes in univariate analysis and not at all in multivariate analyses. Moreover, when MaxSTPre was forced into the multivariate analysis of MaxSTPost, MaxSTPost remained a potent, independent predictor of outcomes, whereas MaxSTPre lost prognostic utility. In combination, these data support the premise that absolute STE after procedure is not just a surrogate for the magnitude of myocardium at risk at presentation, but rather a valid index of STR and reperfusion success.
Notably, most previously published data on STR in AMI are from thrombolysis trials, in which TIMI flow grade 3 is restored in 50% to 60% of patients. In this setting, STR has been studied as a means for predicting target vessel patency and thereby stratifying patients for rescue angioplasty. In contrast, TIMI flow grade 3 is restored in most patients after primary PCI, and their prognosis is favorable. Several smaller, single-center retrospective trials have found a relationship between the relative degree of STR after primary PCI and mortality (16,24,25). The present study is the first to demonstrate the independent value of STR from a large, multicenter primary PCI trial. Moreover, the novel demonstration that simple measures such as MaxSTPost and High Risk categorization are at least as prognostically useful as more complex traditional algorithms has immediate applications for further risk stratification and identification of patients who may benefit from additional therapies, such as interventions to enhance microvascular perfusion.
An additional important finding from the present investigation is that the absolute and relative degree of STR is an important independent predictor of reinfarction during the follow-up period. The mechanism underlying the higher rates of reinfarction with incomplete STR is unknown,although it may involve decreased microcirculatory patency (16,25) or the presence of a prothrombotic state in patients with extensive reperfusion injury (26).
As with most prior trials investigating STR, the current report describes a post-hoc analysis and thus should be validated in future prospective studies. Quantitative ST-segment evaluation was not part of the original study protocol, explaining the relatively high rate of exclusion. Thus, strict inclusion criteria were specified for this study with regard to ECG quality, ECG timing, and degree of baseline STE to maintain a homogeneous and high-quality population. Almost one-third of patients were excluded because ECGs were outside our specified time frame. The pre-specified window of up to 4 h after PCI is wider than that used in some prior trials. However, the average time from balloon inflation to post-procedure ECG (1.77 h) was similar to previous studies (6,7,18–20,24). Moreover, the optimal time interval to measure STR after primary PCI has not been determined, and it is unknown whether earlier STR determination carries more prognostic weight. Regardless, the present study demonstrates that STR utilizing the current time window is prognostically meaningful, independent of epicardial vessel patency.
The most significant difference between the substudy cohort and excluded patients was the greater percentage of anterior MIs in the substudy. This is likely due to the fact that infarcts associated with right coronary and left circumflex coronary artery lesions were less likely to achieve our pre-specified level of STE in multiple leads and were more likely to have been assessed asan indeterminate location. Moreover, in the final stage of exclusion, 225 patients (11%) were excluded because of technically inadequate ECGs. The majority of these patients had baseline artifacts in one or more ECG leads that precluded precise assessment of STE throughout the infarct territory. This rate of exclusion, which is common in our core laboratory experience with static ECG tracings, highlights a limitation of traditional complex STR algorithms, which can be confounded by even modest technical inadequacies. Additional final-stage exclusions were also due to transient arrhythmias, ectopy, pacing, or indeterminate infarct location (reflecting the greater number of left circumflex lesions excluded at this stage). Nonetheless, the otherwise similar baseline characteristics and event-free survival rates between patients excluded and included suggest that the bias created was minimal.
This study has established that the simple measure of the maximal residual degree of STE on a single post-procedure ECG is a strong independent predictor of both survival and freedom from reinfarction after primary PCI. Notably, however, despite restoration of TIMI flow grade 3 in 96% of patients, MaxSTPost <1 mm was present in only 59% of patients after PCI. By this measure, reperfusion is suboptimal in 41% of patients after contemporary percutaneous interventional approaches. Future studies evaluating reperfusion strategies should emphasize nontraditional measures of reperfusion success, including STR and myocardial perfusion (16,20,24,27), rather than epicardial flow rates.
- Abbreviations and acronyms
- acute myocardial infarction
- Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications
- left-bundle branch block
- myocardial infarction
- minimal ST-segment deviation
- percutaneous coronary intervention
- ST-segment elevation
- ST-segment elevation resolution
- Thrombolysis In Myocardial Infarction
- Received April 22, 2004.
- Revision received June 9, 2004.
- Accepted June 14, 2004.
- American College of Cardiology Foundation
- von Essen R.,
- Schmidt W.,
- Uebis R,
- et al.
- Saran R.K.,
- Been M.,
- Furniss S.S.,
- Hawkins T.,
- Reid D.S.
- Blanke H.,
- Scherff F.,
- Karsch K.R,
- et al.
- Schroder R.,
- Dissmann R.,
- Bruggemann T,
- et al.
- Schroder R.,
- Wegscheider K.,
- Schroder K.,
- Dissmann R.,
- Meyer-Sabellek W.
- Buszman P.,
- Szafranek A.,
- Kalarus Z.,
- Gasior M.
- Matetzky S.,
- Novikov M.,
- Gruberg L,
- et al.
- Shah A.,
- Wagner G.S.,
- Granger C.B,
- et al.
- Poli A.,
- Fetiveau R.,
- Vandoni P,
- et al.
- Mauri F.,
- Maggioni A.P.,
- Franzosi M.G,
- et al.
- van't Hof A.W.,
- Liem A.,
- Suryapranata H,
- et al.
- Claeys M.J.,
- Bosmans J.,
- Veenstra L,
- et al.
- Okada H.,
- Woodcock-Mitchell J.,
- Mitchell J,
- et al.
- Stone G.W.,
- Peterson M.,
- Lansky A.J,
- et al.