Author + information
- Received June 15, 2005
- Revision received July 25, 2005
- Accepted August 20, 2005
- Published online March 7, 2006.
- Amir Halkin, MD⁎,
- Gregg W. Stone, MD, FACC†,⁎ (, )
- Cindy L. Grines, MD, FACC‡,
- David A. Cox, MD, FACC§,
- Barry D. Rutherford, MD, FACC∥,
- Paolo Esente, MD, FACC¶,
- Carol M. Meils, MD, FACC#,
- Per Albertsson, MD, FACC⁎⁎,
- Anthony Farah, MD, FACC††,
- James E. Tcheng, MD, FACC‡‡,
- Alexandra J. Lansky, MD, FACC⁎,† and
- Roxana Mehran, MD, FACC⁎,†
- ↵⁎Reprint requests and correspondence:
Dr. Gregg W. Stone, The Cardiovascular Research Foundation, 55 East 59th Street, 6th Floor, New York, New York 10022
Objectives We examined the prognostic implications of the absolute level and rate of increase of creatine kinase (CK) elevation after primary percutaneous coronary intervention (PCI).
Background Peak creatine kinase (CKpeak) and the rate of CK increase are related to reperfusion success and clinical outcomes after thrombolytic therapy for acute myocardial infarction (AMI). The utility of routine serial CK monitoring after primary PCI, in which normal antegrade blood flow is restored in most patients, is unknown.
Methods In the Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) trial, 1,529 patients with AMI randomized to either stenting or balloon angioplasty, each with or without abciximab, had CK levels determined at baseline and at 8 ± 1 h, 16 ± 1 h, and 24 ± 1 h after PCI.
Results The CKpeakoccurred at baseline in 3.9% of patients, at 8 ± 1 h in 69.6%, at 16 ± 1 h in 20.0%, and at 24 ± 1 h in 6.5%. The CK levels at all post-procedural time points were significantly higher in patients who died compared with the one-year survivors, as was CKpeak(mean, 2,865 U/l vs. 1,885 U/l, respectively, p ≤ 0.001). By multivariate analysis, CKpeakwas a significant predictor of one-year mortality (hazard ratio = 2.15, p = 0.0002), independent from post-PCI Thrombolysis In Myocardial Infarction (TIMI) flow. Both the improvement in left ventricular ejection fraction from baseline to seven months and its absolute level were inversely correlated with CKpeak(p < 0.001 for both). In contrast, the time to CKpeakwas not an independent predictor of mortality or myocardial recovery.
Conclusions The CKpeakafter primary PCI is a powerful predictor of one-year mortality independent of other clinical and angiographic measures.
In the pre-reperfusion era, the peak level of serum creatine kinase (CK) measured within the first 24 to 48 h after acute myocardial infarction (AMI), as a reflection of infarct size, was found to be strongly related to subsequent mortality (1). However, the prognostic significance of peak cardiac enzyme level elevation after reperfusion therapy is less clear. Early reperfusion therapy during evolving AMI is known to increase the absolute peak CK level and the rapidity with which the apex of the peak is reached (2). Although a rapid increase in CK is a marker of effective thrombolysis (3,4), some studies indicate that larger (5) and earlier (6) enzyme peaks are associated with increased rates of adverse outcomes after pharmacologic reperfusion therapy. In recent years, primary percutaneous coronary intervention (PCI) has emerged as the preferred reperfusion modality for AMI (7). Whether CK elevations have similar prognostic implications after primary PCI as thrombolytic therapy, given the higher rates of Thrombolysis In Myocardial Infarction (TIMI) flow grade 3 and reperfusion success with the catheter-based strategy, is unknown and has not been studied.
We therefore performed a post hoc analysis of the database from a large, prospective, multicenter, randomized study of mechanical reperfusion therapies in AMI to determine the relationship between post-PCI CK levels, kinetics, and outcomes.
Details of the Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications (CADILLAC) trial have been previously reported (8). Briefly, 2,082 patients ≥18 years of age with AMI and symptom onset within 12 h undergoing primary PCI were randomized to primary balloon angioplasty versus stent implantation, each with or without abciximab. The principal clinical exclusion criteria were cardiogenic shock and the prior use of thrombolytic therapy for the present admission. Angiographic inclusion criteria included a native coronary artery culprit vessel with a reference diameter of 2.5 to 4.0 mm and a lesion length ≤64 mm. Clinical follow-up continued for one year. Protocol-specified, angiographic 7-month follow-up was completed in 656 patients (73% of a 900-patient pre-specified cohort). Quantitative coronary angiography and ventriculographic assessment were performed using dedicated software (QCA-CMS, MEDIS, Leiden, the Netherlands) at an independent core angiographic laboratory at the Cardiovascular Research Foundation, New York, New York. Left ventricular ejection fraction (LVEF) was calculated using the length-area method (9). Antegrade coronary blood flow was evaluated using the TIMI scale (10). Myocardial blush (11) and changes in electrocardiographic ST-segment elevation (12) were determined in two formal CADILLAC sub-studies consisting of 1,301 and 700 patients, respectively.
Protocol-specified blood sampling for CK levels was performed at baseline and at 8 ± 1 h, 16 ± 1 h, and 24 ± 1 h after PCI. Measurement of serum total CK levels was performed according to local hospital standards. Outcomes were examined stratified by peak CK levels (CKpeak), divided into four equal quartiles.
End points and statistical analysis
The primary study end point was a composite of major adverse cardiac events, defined as death from any cause, reinfarction, repeat target vessel revascularization as a result of ischemia, or disabling stroke. The components of the composite end point have been previously defined (8). Outcomes were evaluated as a function of absolute CKpeaklevel (as both a continuous and a categorical [quartile] measure), and by the timing of CKpeak. Categorical data were compared using the Fisher exact test. Continuous variables are presented as mean ± SD and were compared using the Student ttest or one-way analysis of variance as appropriate. For comparison of multiple groups with ordinal categorization, the trend in the binomial proportions of categorical variables was analyzed using the Cochran-Armitage test, and analysis of variance was used to test for linear trends across the means of continuous variables. Clinical outcomes were presented as Kaplan-Meier survival estimates and were compared using the log-rank test. Independent predictors of CKpeakwere determined by multiple linear regression using stepwise selection (entry and exit criteria of p < 0.10 and p < 0.15, respectively), considering CKpeakas a log-transformed continuous variable. Independent predictors of survival were determined with Cox proportional hazard regression analysis using stepwise selection of all of the parameters in Table 1with entry and exit criteria of p < 0.10 and p < 0.15, respectively. For all analyses, a two-tailed p < 0.05 was considered statistically significant.
Of 2,082 patients, CK values were available at baseline and at all three post-PCI pre-specified time points in 1,529 patients (73.4%), who are the study cohort. Baseline characteristics and procedural results were similar in patients with compared to those without complete CK data (Table 2).
Clinical characteristics and procedural results of patients stratified by CKpeakare shown in Table 1. Patients with higher CKpeakvalues were more likely to be younger, male, and non-diabetic, and to have a shorter time from symptom onset to PCI than those with lower CKpeakvalues. The infarct artery of patients with a greater CKpeakvalue was larger in diameter and most commonly the left anterior descending vessel, and baseline global and regional left ventricular function was depressed. Spontaneous reperfusion before intervention was also less frequent in patients with a greater CKpeakvalue, and the post-procedure rates of normalized TIMI flow grade and myocardial blush were lower than in patients with a greater CKpeakvalue.
Pattern and determinants of CK elevation
For the entire study population, the mean CKpeakwas 1,921 ± 1,723 U/l. By multivariate analysis, the independent correlates of CKpeakwere left anterior descending culprit artery (odds ratio [95% confidence interval] = 1.90 [1.43 to 2.52], p < 0.0001), pre-procedural TIMI flow grade 0 to 2 (5.26 [3.23 to 9.09], p < 0.0001), younger age (1.02 [1.01 to 1.03], p < 0.0001), and reduced baseline LVEF (1.05 [1.04 to 1.06], p < 0.0001). Of the variables listed in Table 1, none of the post-procedural angiographic findings were independently predictive of CKpeak.
The mean CK level at each time period for patients within each CKpeakquartile are shown in the upper panel of Figure 1.The CKpeakoccurred at the baseline measure in 3.9% of patients, at 8 ± 1 h in 69.6% of patients, at 16 ± 1 h in 20.0% of patients, and at 24 ± 1 h in 6.5% of patients. As seen in the lower panel of Figure 1, the greater the CKpeak, the earlier the time the peak was achieved.
The CK levels at all time points as well as at CKpeakwere similar in patients stratified by randomization arm (Fig. 2).Neither stent implantation nor abciximab use were independent predictors of CKpeakby multivariate analysis.
The one-year all-cause mortality was 3.8%. Although similar at baseline, CK levels at all post-procedural time points (Fig. 3,top panel) and CKpeak(2,865 ± 2,158 U/l vs. 1,885 ± 1,694 U/l, respectively, p = 0.0001) were significantly higher in patients who died compared with those who survived to one-year follow-up. Similarly, the mean CK level at each time point (Fig. 3, bottom panel) and CKpeak(2,866 ± 2,191 U/l vs. 1,846 ± 1,664 U/l, respectively, p = 0.001) in those who died versus in one-year survivors were significant predictors of one-year mortality when confined to patients in whom TIMI flow grade 3 was restored after PCI.
Clinical and angiographic outcomes in patients stratified by CKpeakas a categorical variable are shown in Table 3and in Figure 4.Mortality at 30 days and at 1 year increased significantly with an increasing CKpeaklevel. There was a significant inverse relationship between CKpeakand both the absolute LVEF at seven months and its improvement from baseline to follow-up (Table 3, Fig. 5).No relationship was present between CKpeakand the rates of reinfarction, stroke, or target vessel revascularization (Table 3).
Because in 89.6% of the study population CKpeakoccurred within 8 to 16 h after PCI, a secondary analysis examined 30-day and 1-year survival among the 1,871 patients (89.9% of the entire CADILLAC study population) with at least one available CK measurement at either of those time points. When stratified by CKpeakquartiles, mortality increased significantly with increasing CKpeak(30-day mortality in quartiles 1 through 4: 0.4%, 0.6%, 1.1%, 4.5%, respectively; p < 0.0001; 1-year mortality: 2.2%, 2.2%, 3.7%, 7.6%, respectively; p < 0.0001).
CKpeakkinetics and outcomes
Baseline features and the outcomes of patients stratified by the timing of CKpeakare shown in Table 4.Among the 96.1% of patients in whom CK peaked after PCI, a more rapid time to CKpeakwas associated with younger age, infarction of the left anterior descending artery, less frequent spontaneous reperfusion before PCI, abciximab use, a shorter time from symptom onset to PCI, and a greater CKpeakvalue. A faster time to CKpeakwas also associated with reduced LVEF at baseline and at follow-up, although the extent of improvement in LVEF from baseline to follow-up was independent of time to CKpeak. There was also no significant relationship between the rapidity to CKpeakon mortality at 30 days or 1 year.
In comparison with patients in whom the CKpeakoccurred after PCI, the 3.9% of patients in whom the CKpeakhad already occurred at baseline had a longer time from symptom onset to PCI, were less likely to have TIMI flow grade 3 at baseline, and had smaller infarct vessels with less frequent stent implantation. Despite a lower mean CKpeakvalue, patients with CKpeakat baseline had greater mortality at 30 days and at 1 year compared with patients with CKpeakat 8 or 16 h.
Multivariate predictors of mortality
By Cox proportional hazard regression analysis, CKpeakas a continuous variable, but not the timing of CKpeak, was an independent predictor of one-year mortality (Table 5).When CKpeakwas entered into the multivariate model as a categorical variable, the hazard ratios and 95% confidence intervals for one-year mortality with quartile 1 (CKpeak<669 U/l) as reference (HR = 1.00) were: for quartile 2 (669 ≤ CKpeak<1,458 U/l), HR = 1.77 [0.42 to 7.42], p = 0.44; for quartile 3 (1,458 < CKpeak≤2,667 U/l), HR = 2.97 [0.81 to 10.86], p = 0.10; and for quartile 4 (CKpeak>2,667 U/l), HR = 5.63 [1.61 to 19.62], p = 0.007.
In a secondary analysis, CKpeakas a continuous variable remained an independent correlate of one-year survival when the multivariate model was applied to the 1,871 patients with an available CK measurement at 8 to 16 h after PCI (HR = 2.52 [1.58 to 4.03], p = 0.0001).
The principal findings of the present study are: 1) despite achievement of TIMI flow grade 3 in >90% of patients, CKpeakafter primary PCI is a powerful independent determinate of one-year mortality; and 2) the CKpeakafter primary PCI is also inversely related to recovery of left ventricular function.
Before the advent of reperfusion therapy for AMI, the peak level of myocardial enzymes was shown to correlate with infarct size and survival (1,13,14). More recently, the Global Utilization of Streptokinase and TPA for Occluded Coronary Arteries (GUSTO)-IIb investigators reported that the extent of CK elevation is an independent predictor of mortality in a wide spectrum of patients with acute coronary syndromes, including those treated with reperfusion therapy (15). However, experimental studies have suggested that enzymatic measures after early reperfusion overestimate infarct size (16,17). Furthermore, in recent reports higher enzyme peaks after thrombolysis predicted a lower LVEF and an increased risk of congestive heart failure and death (5,18). Prior studies have also reached conflicting conclusions regarding the prognostic implications of the time to CKpeakafter thrombolysis. An early CKpeakafter thrombolysis has been associated with successful reperfusion in some studies (3,4), but with a larger infarct size in others (6). In the present study as well, a higher CKpeakwas associated with increased 30-day and 1-year mortality. Unlike thrombolytic studies in which flow in the infarct artery before reperfusion therapy is not known, in the present study we were able to ascertain that a higher CKpeakwas associated with reduced rates of spontaneous reperfusion before angioplasty. Given the beneficial prognostic implications of normalized flow in the infarct artery at presentation, it is plausible that this factor is an important determinant of myocardial salvage in the setting of primary angioplasty.
TIMI flow grade 3 is present in <60% of patients after thrombolysis; this observation may explain some of the heterogeneity of findings in the aforementioned studies. In this regard it is noteworthy that no large-scale study has examined the implications of the absolute level and rate of increase of CK after primary PCI, in which TIMI flow grade 3 may be restored in >90% of patients. Biomarkers other than CK, however, have been examined in the setting of primary PCI. The cumulative extent of lactate dehydrogenase release after primary PCI was found to be an independent predictor of one-year mortality in a single study (19). However, this enzyme, which does not peak until ∼72 h, is not a convenient bedside tool and is prognostically ineffective in patients who die within the first few days after AMI, the highest-risk period for mortality (20). For these reasons, the use of lactate dehydrogenase for the diagnosis of AMI is no longer endorsed (21). In contrast, serial measures of CK release are universally obtained as standard of care after AMI, and peak typically within 24 h. As such, characterizing the prognostic implications of CK release after primary PCI is of potential clinical utility.
CK elevation and outcomes after primary PCI
The present analysis shows that peak CK level is a powerful predictor of one-year mortality after primary PCI, independent of post-procedure TIMI flow grade. Although the mechanistic link between CK levels and survival is uncertain, the significant inverse association between CKpeakand the absolute improvement in LVEF from baseline to seven-month follow-up suggests that the increase in CK after primary PCI is a reflection of infarct size.
Unlike CKpeak, and in contrast to prior thrombolytic trials (3,4), an earlier CK increase in the CADILLAC trial was not associated with angiographic markers of enhanced reperfusion such as higher rates of TIMI flow grade 3 and normal myocardial blush. Conversely, an early post-PCI CKpeakwas associated with higher maximum CK levels and a lower LVEF at baseline and at follow-up, in accordance with the findings of others (6). Whether the association between a rapidly increasing CK and reduced left ventricular function results from more rapid enzyme release by larger infarcts or whether it is an expression of reperfusion injury (22,23) is unknown. An association was also observed between CKpeakkinetics and survival, with a trend toward higher mortality present in patients with later time to peak enzyme increase, despite lower CKpeak. However, by multivariate analysis only CKpeak, and not time to peak, was an independent predictor of survival. Finally, the small cohort (3.9%) of patients in whom the CKpeakwas reached at baseline (before PCI) had a different clinical and angiographic profile than the rest, with longer times to reperfusion and smaller infarct arteries. It is likely that the true CKpeakmay have occurred even earlier, possibly explaining the trend toward higher mortality in this group despite higher rates of pre-procedural spontaneous reperfusion (24).
Correlates of peak CK elevation
The strongest correlates of CKpeakwere left anterior descending culprit artery, pre-procedural TIMI flow grade 0 to 2, younger age, and reduced baseline LVEF. The greater CKpeakin left anterior descending artery occlusion may be explained not only by the larger myocardial territory supplied, but also by decreased reperfusion success in this territory (12). The fact that lack of spontaneous reperfusion in the infarct vessel before PCI was a multivariate predictor of CKpeakadds further credence to the concept of pharmacologic facilitation of primary PCI to restore reperfusion as early as possible before definitive mechanical revascularization (25), a strategy currently undergoing clinical investigation. The correlation between reduced baseline LVEF and CKpeaklikely reflects a greater amount of myocardium at risk and highlights the prognostic utility of assessing left ventricular function during the index procedure. The lower CKpeakin elderly patients and in those with prior MI might be explained by reduction of myocardial mass and, in diabetic patients, by lower myocardial expression of CK (26,27). Of note, abciximab use and stenting were not correlates of CKpeakin the CADILLAC trial, consistent with previous smaller studies (28,29). This observation is also consistent with the prior reported absence of a relationship between these modalities and recovery of left ventricular function (30).
Approximately 26% of patients were excluded from the present study because of missing CK values at one or more time points. Nevertheless, these patients did not differ from those included with respect to important clinical and angiographic features. Interlaboratory variability in the determination of serum CK concentration must be recognized (31), and the findings of this study should not be misconstrued to suggest that a specific CK level, or cutoff, can be used for the purposes of prognostication or selection of treatment strategies. It is also worth noting that patients with some conditions that influence CK levels were either excluded or underrepresented in the CADILLAC trial (e.g., severe renal failure, rhabdomyolysis), and that the use of more specific biomarkers of myocardial necrosis (e.g., CK-MB isoenzymes or troponins) rather than total CK may add further prognostic accuracy. Also, it is possible that a marker detectable sooner after AMI onset than CK (e.g., myoglobin) might provide incremental prognostic information (2). These limitations do not detract from the powerful clinical utility of post-PCI CK evaluation, an inexpensive test that is more readily available in certain health care systems than assays for alternative cardiac biomarkers (32). The implications of CK increase in terms of left ventricular functional recovery should be considered within the context of the timing of protocol-defined determination of convalescent LVEF at seven-month follow-up and attrition of the study population, which itself was impacted by CKpeak. Protocol-specified blood sampling every eight hours in the present study did not permit accurate calculations of the area under the CK time-activity curve (33), a potentially more accurate measure of infarct size determination and prognosis. However, CK sampling as performed in this study closely reflects routine clinical practice, and CKpeakis a simpler clinical measure than time-activity curves. The absolute CKpeakand time to CKpeakmay have been underestimated in the 6.5% of patients in whom CKpeakoccurred at 24 h, the late peak possibly reflecting unsuccessful myocardial reperfusion. Finally, the applicability of our findings to patient populations dissimilar to that studied in the CADILLAC trial or in which alternative strategies for the management of AMI were used is unknown.
Conclusions and clinical implications
Independent of other clinical and angiographic measures, CKpeakdirectly correlates with infarct size and long-term mortality in patients with AMI treated with primary PCI, and as such, the assessment of serial CK levels in the first 24 h after admission continues to be of prognostic utility in the contemporary reperfusion era. Using CKpeak, 25% of patients will be identified within 24 h of hospital admission in whom the hazard for one-year mortality is approximately five- to six-fold increased, signifying a patient cohort in whom intensive monitoring and secondary preventative measures are warranted. Despite successful primary PCI, novel strategies must also be identified to further enhance myocardial salvage and prolong survival in this high-risk group.
Funding for the CADILLAC trial was provided by Guidant Corporation.
- Abbreviations and Acronyms
- acute myocardial infarction
- Controlled Abciximab and Device Investigation to Lower Late Angioplasty Complications
- creatine kinase
- peak creatine kinase
- left ventricular ejection fraction
- percutaneous coronary intervention
- Thrombolysis In Myocardial Infarction
- Received June 15, 2005.
- Revision received July 25, 2005.
- Accepted August 20, 2005.
- American College of Cardiology Foundation
- Thompson P.L.,
- Fletcher E.E.,
- Katavatis V.
- Zabel M.,
- Hohnloser S.H.,
- Koster W.,
- Prinz M.,
- Kasper W.,
- Just H.
- Costantini C.O.,
- Stone G.W.,
- Mehran R.,
- et al.
- McLaughlin M.G.,
- Stone G.W.,
- Aymong E.,
- et al.
- Sobel B.E.,
- Bresnahan G.F.,
- Shell W.E.,
- Yoder R.D.
- Hackel D.B.,
- Reimer K.A.,
- Ideker R.E.,
- et al.
- Savonitto S.,
- Granger C.B.,
- Ardissino D.,
- et al.
- Tamaki S.,
- Murakami T.,
- Kadota K.,
- et al.
- Elsman P.,
- Zijlstra F.,
- Miedema K.,
- et al.
- Alpert J.S.,
- Thygesen K.,
- Antman E.,
- Bassand J.P.
- Hansen P.R.
- Rentrop K.P.,
- Feit F.,
- Sherman W.,
- Thornton J.C.
- Stone G.W.,
- Cox D.,
- Garcia E.,
- et al.
- Popovich B.K.,
- Boheler K.R.,
- Dillmann W.H.
- Awaji Y.,
- Hashimoto H.,
- Matsui Y.,
- et al.
- Brener S.J.,
- Barr L.A.,
- Burchenal J.E.,
- Wolski K.E.,
- Effron M.B.,
- Topol E.J.
- Tcheng J.E.,
- Kandzari D.E.,
- Grines C.L.,
- et al.
- Henderson A.R.,
- Krishnan S.,
- Webb S.,
- Cheung C.M.,
- Nazir D.J.,
- Richardson H.
- Hasdai D.,
- Behar S.,
- Boyko V.,
- Danchin N.,
- Bassand J.P.,
- Battler A.
- Roberts R.