This article has a correction. Please see:
Author + information
- Received February 9, 2001
- Revision received May 29, 2001
- Accepted June 15, 2001
- Published online October 1, 2001.
- Karen Klatte, MD∗,
- Bernard R. Chaitman, MD, FACC∗,2,* (, )
- Pierre Theroux, MD, FACC†,2,
- Jeffrey A. Gavard, PhD∗,
- Karen Stocke, MBA∗,
- Steven Boyce, MD, FACC‡,2,
- Claus Bartels, MD§,2,
- Birgit Keller, MS∥,2,
- Andreas Jessel, MD∥,2,
- for the GUARDIAN Investigators1
- ↵*Reprint requests and correspondence: Dr. Bernard R. Chaitman, St. Louis University Health Sciences Center, Division of Cardiology (15th Floor), 3635 Vista Avenue at Grand Boulevard, P.O. Box 15250, St. Louis, Missouri 63110-0250
We sought to determine if elevated cardiac serum biomarkers after coronary artery bypass graft surgery (CABG) are associated with increased medium-term mortality and to identify patients that may benefit from better postoperative myocardial protection.
The relationship between the magnitude of cardiac serum protein elevation and subsequent mortality after CABG is not well defined, partly because of the lack of large, prospectively studied patient cohorts in whom postoperative elevations of cardiac serum markers have been correlated to medium- and long-term mortality.
The GUARD during Ischemia Against Necrosis (GUARDIAN) study enrolled 2,918 patients assigned to the entry category of CABG and considered as high risk for myocardial necrosis. Creatine kinase-myocardial band (CK-MB) isoenzyme measurements were obtained at baseline and at 8, 12, 16 and 24 h after CABG.
The unadjusted six-month mortality rates were 3.4%, 5.8%, 7.8% and 20.2% for patients with a postoperative peak CK-MB ratio (peak CK-MB value/upper limits of normal [ULN] for laboratory test) of <5, ≥5 to <10, ≥10 to <20 and ≥20 ULN, respectively (p < 0.0001). The relationship remained statistically significant after adjustment for ejection fraction, congestive heart failure, cerebrovascular disease, peripheral vascular disease, cardiac arrhythmias and the method of cardioplegia delivery. Receiver operating characteristic curve analysis revealed an area under the curve of 0.648 (p < 0.001); the optimal cut-point to predict six-month mortality ranged from 5 to 10 ULN.
Progressive elevation of the CK-MB ratio in clinically high-risk patients is associated with significant elevations of medium-term mortality after CABG. Strategies to afford myocardial protection both during CABG and in the postoperative phase may serve to improve the clinical outcome.
The relationship between the magnitude of creatine kinase-myocardial band (CK-MB) isoenzyme elevation and subsequent mortality after coronary artery bypass graft surgery (CABG) is not well defined because of the absence of large, prospectively studied patient cohorts in whom post-procedural elevations of CK-MB have been correlated to medium- and long-term mortality (1–5). Most patients have some degree of cardiac protein elevation after CABG that can be detected using specific CK-MB bioassays (6). A range of CK-MB 5 to 10 times the upper limits of normal (ULN) is often used in clinical trials to identify a myocardial infarction (MI) after CABG (1,7). However, there is sparse data examining whether the magnitude of CK-MB elevation after CABG correlates with future mortality.
The recent joint European Society of Cardiology and American College of Cardiology consensus document on the re-definition of MI states that any degree of myocardial necrosis caused by myocardial ischemia and detected using cardiac serum markers after a revascularization intervention should be labeled as an MI (8). The report recommends that more data are needed to determine the magnitude of risk for elevated cardiac serum markers after CABG.
The primary aim of this study was to determine whether postoperative CK-MB elevation and electrocardiographic (ECG) changes are associated with adverse six-month mortality after CABG. The hypothesis was tested in 2,918 patients entered into the CABG arm of the GUARD during Ischemia Against Necrosis (GUARDIAN) study (9,10).
The GUARDIAN study was a multicenter, multinational, double-blinded, placebo-controlled, randomized clinical trial that examined the efficacy of the Na+/H+exchange inhibitor cariporide (HOE-642A) on the reduction of death or MI (9,10). Patient recruitment occurred from May 1997 to August 1998. This study was approved by the Institutional Review Board at each participating institution, and written, informed consent was obtained from all patients. The patients’ inclusion and exclusion criteria are described elsewhere; patients at risk of myocardial necrosis due to unstable angina or non–Q-wave MI (n = 5,233), high-risk percutaneous transluminal coronary angioplasty (PTCA) (n = 3,439) or high-risk CABG (n = 2,918) were included (9). Patients were deemed to be at high risk for CABG if they required an urgent intervention (such as after failed PTCA) or repeat CABG, or if they had a history of angina at rest or at minimal exercise within the past four weeks and with the presence of two or more of the following: age >65 years, female gender, diabetes mellitus, left ventricular ejection fraction <35% and left main or three-vessel coronary artery disease. Patients with new ST segment elevation MI, significant hepatic or renal impairment or secondary causes of unstable angina, such as anemia or hyperthyroidism, were excluded. The protocol specified that all patients be contacted at six months to ascertain their vital status, occurrence of MI and re-admission to the hospital.
The patients’ demographic information, medical/surgical history, 12-lead ECG and CK-MB enzyme levels were collected at baseline, before CABG. The CK-MB enzyme levels were also collected at 8, 12, 16 and 24 h after CABG. Absolute units of CK-MB enzymes were graded on a four-point scale (≤60, >60 to ≤100, >100 to ≤200 and >200 U/l). The peak CK-MB ratio was determined by dividing the peak CK-MB value by the ULN for that laboratory, because the ULN for CK-MB and type of assay used varied across laboratories. Four categories were examined (<5, ≥5 to <10, ≥10 to <20 and ≥20 ULN).
Electrocardiograms obtained within 14 days postoperatively were considered for analysis. The ECGs were classified according to the Minnesota Code using an adaptation of the Novacode system for serial Q-wave comparisons (11,12). Serial ECG abnormalities from baseline were coded as one- and two-step incremental changes of worsening severity, as previously described (13).
Of the 2,918 patients initially entered into the CABG group of GUARDIAN, six-month vital status was known for 2,891 (99.1%). Patients were excluded if they died within 24 h of CABG (n = 24), had MI at study entry (preoperative CK-MB ratio > ULN) (n = 96), had no preoperative CK-MB enzyme determination (n = 137) or had an insufficient number of postoperative CK-MB values available to classify the event (n = 111), or if concomitant valve replacement surgery was performed (n = 129). Patients with concomitant valve surgery were excluded because their six-month mortality rate was 17.8% vs. 5.5% for patients without concomitant valve surgery, p < 0.001), and their results represented only 4.4% of the total group retained for analysis. The study population consisted of the remaining 2,394 patients who underwent only CABG, 131 of whom died during the six-month follow-up period.
Differences in baseline demographic data, medical/surgical histories and health characteristics between patients who died during follow-up and those who survived were assessed using the chi-square test for categorical variables. The nonparametric Kolmogorov-Smirnov test was used for all continuous variables (14). Odds ratios and relative risks for six-month mortality were calculated for CK-MB and ECG categories in relation to the baseline level in that category (<5 ULN, ≤60 absolute units or <2 steps of Minnesota Code worsening). Mortality was expressed per 100 patients (%). Cumulative six-month survival by CK-MB categories was performed using life-table analysis according to 10-day increments after CABG (15). The independent ability to predict six-month mortality using demographic data, medical/surgical history, CK-MB enzyme levels and ECG variables was assessed by multiple logistic regression analysis (16,17). Receiver operating characteristic (ROC) curve analysis was performed to determine the prognostic performance of different cut-points for the CK-MB ratio to six-month mortality. All statistical analyses were performed using SPSS, version 10.0 for Windows (SPSS, Chicago, Illinois) and EPI Info, version 6.04b (CDCP, Atlanta, Georgia).
Demographic data and medical/surgical history
Patients who died during the six-month follow-up had a lower baseline ejection fraction and were more likely to have had a history of MI, congestive heart failure, cerebrovascular disease, peripheral vascular disease, cardiac arrhythmias and insulin-dependent diabetes mellitus, compared with patients who survived. Retrograde cardioplegia delivery was less likely to be associated with six-month mortality. Intermittent cross-clamping was more likely to be associated with six-month mortality. Most patients received a cold cardioplegic solution. The number and type of graft(s) inserted, the use of blood versus crystalloid solution and the temperature of the cardioplegic solution were not significantly different between survivors and nonsurvivors.
Relationship between peak CK-MB enzyme levels and six-month mortality
Increasing peak CK-MB enzyme ratios were associated with a greater risk of death (Table 2). The six-month mortality rates for patients with peak CK-MB ratios of <5, ≥5 to <10, ≥10 to <20 and ≥20 ULN were 3.4%, 5.8%, 7.8% and 20.2%, respectively (p < 0.0001). The test for trend was significant (chi-square = 70.52 at 1 df, p < 0.0001). The converse is shown in Figure 1, where cumulative six-month survival was inversely related to the peak CK-MB ratio (p < 0.0001). Similar results were obtained when the patients were stratified by peak CK-MB enzyme values (p < 0.0001) (Table 3, Fig. 2). The area under the curve (AUC) on ROC analysis was 0.648 (p < 0.0001) (Fig. 3). The sensitivity and specificity for using various cut-points of peak CK-MB enzyme ratios to correctly identify patients who did and did not die during the six-month follow-up period indicate that the optimal cut-point for prognostic performance lies between 5 and 10 ULN. The sensitivity and specificity for the 5-, 10-, 15- and 20-ULN cut-points were 0.60 and 0.66, 0.39 and 0.85, 0.29 and 0.91 and 0.26 and 0.94, respectively.
Relationship between ECG abnormalities and six-month mortality
The presence of new Q-wave worsening of two grades, according to the Minnesota Code, was significantly associated with six-month mortality. One hundred and ten patients developed new Q-wave worsening of two grades (Minnesota Code) after CABG. The six-month mortality rate was 10.9% in these 110 patients, compared with 4.8% in the 2,222 patients without new postoperative Q-waves (p < 0.01). Thirty-eight of the 110 patients had peak CK-MB ratios <5 ULN, and 72 had peak CK-MB ratios ≥5 ULN. The six-month mortality rates were 2.6% and 15.3%, respectively (p = 0.09). The six-month mortality rates for the 2,222 patients who did not develop a new Q-wave worsening of two grades (Minnesota Code) were 3.1% for patients whose peak CK-MB ratio was <5 ULN and 8.0% for patients whose peak CK-MB ratio was ≥5 ULN (p < 0.0001).
Multivariate modeling for six-month mortality risk
The independent ability to predict six-month mortality using all significant univariate variables was examined through multiple logistic regression analysis. The results of the final model are presented. The strongest predictor of six-month mortality was the peak CK-MB enzyme ratio. When the model was run as a binary variable (≥5 vs. <5 ULN), two-grade Q-wave worsening (Minnesota Code) was a significant independent predictor of six-month mortality (odds ratio 1.99, p < 0.05). Two models were run to adjust for the 751 patients with missing ejection fraction measurements. The first model imputed the mean ejection fraction for all patients with missing values; the second model used a category of “missing” for patients with a missing ejection fraction. The models gave virtually identical results. The findings were similar when peak CK-MB value categories, rather than peak CK-MB ratio categories, were tested in the model.
Of the 131 patients who died during the six-month follow-up period after CABG, 43 occurred within 10 days of the operation. To examine the impact of these early deaths on the relationship between the peak CK-MB ratio and six-month mortality, the analyses were repeated using only patients that survived >10 days after CABG (Table 6). A history of cerebrovascular disease was no longer predictive of six-month mortality; previous PTCA and the number of bypass grafts were new variables that were predictive of decreased six-month mortality. The peak CK-MB ratio remained a significant predictor of six-month mortality in the final multiple logistic regression model, indicating that this finding is representative of the survival experience throughout the duration of the follow-up phase.
This report demonstrates that a large, prospectively studied, clinically high-risk patient population undergoing CABG in 1997 to 1998 has an increasing risk of six-month mortality, with increasing levels of postoperative CK-MB elevation. The risk gradient remained significant after adjustment for ejection fraction, congestive heart failure, cerebrovascular disease, peripheral vascular disease, cardiac arrhythmias and the method of cardioplegia delivery.
The study population in this report consisted of 2,918 patients who were initially entered into the CABG arm of the GUARDIAN study (9,10). The data set is unique in several aspects. The study cohort contained a large number of patients receiving isolated CABG procedures, allowing for a robust analysis of the mortality end point. Cardiac enzyme values were prospectively collected as part of the study protocol, thus avoiding sampling bias. Six-month follow-up was vigorously pursued for the mortality end point, with <1% of patients lost to follow-up. The patients received CABG in the modern era (1997–98) and in a multicenter trial, thus increasing the generalizability of the results. The six-month mortality gradient was observed with increasing levels of peak CK-MB enzyme ratios or with peak absolute units of CK-MB enzymes. The consistency of the data—that is, the strong dose-response relationship between the magnitude of postoperative CK-MB elevation and six-month mortality and the results of multivariate model analysis—strongly argues against these findings being the result of chance. Our data indicate that the relationship between the CK-MB ratio and six-month mortality cannot be entirely attributed to patients who die within the hospital shortly after the operation, because the relationship between the CK-MB ratio and six-month mortality remained significant, even after early deaths (<11 days) were excluded. The likelihood that the elevated postoperative CK-MB values represent myocardial necrosis in most patients is supported by the findings of Force et al. (18), who report an increased frequency of new postoperative regional wall motion abnormalities on two-dimensional echocardiograms, with increasing levels of CK-MB.
Optimal CK-MB ratio cut-point to define MI
Death and MI are a common composite end point used to determine the sample size for clinical trials. Larger MIs, as defined by CK-MB >10 ULN, are a “bad” outcome, and prevention of this magnitude of damage is a desired treatment outcome. Thus, CK-MB elevation ≥10 ULN or >100 U/l within 24 h after CABG constitutes a robust definition of peri-procedural MI that could be used as part of a composite clinical end point of death or MI in studies testing the efficacy of new cardioprotective compounds. The decision to use peak CK-MB enzyme ratios ≥5 or ≥10 ULN to define peri-procedural MI in the setting of clinical studies of CABG depends on the question being addressed (19–21). A more sensitive definition may be desirable to satisfy sample-size requirements in the proof-of-concept phase. The lower cut-point of the peak CK-MB ratio of ≥5 would result in a greater number of end point events (∼1½ times as many for ≥10 ULN). A more stringent and specific definition of a peak CK-MB ratio ≥10 ULN, however, may be preferred to confirm clinical efficacy. The use of a lower cut-point increases the risk of confounding the MI end point variable by including patients with nonischemic, surgically induced myocardial injury. This would affect the specificity of the definition with respect to the medium-term clinical outcome.
Development of a new Q-wave worsening of two grades, according to the Minnesota Code, after CABG was a statistically significant univariate predictor of six-month mortality. After adjustment for baseline variables and postoperative peak CK-MB ratios, however, the risk was attenuated. The adverse long-term (i.e., five years) outcome for patients who develop this new two-grade Q-wave worsening has been well demonstrated in the Coronary Artery Surgery Study and the Bypass Angioplasty Revascularization Investigation (22–24). Major Q-wave items have been observed after CABG by unmasking a previous MI in a small number of patients (25). In GUARDIAN, the relative six-month mortality risk for new postoperative Q-waves was not increased when the peak CK-MB ratio was <5 ULN. Thus, the likelihood of peri-operative myocardial necrosis is greatest when new postoperative Q-waves are accompanied by abnormally elevated cardiac serum biomarkers, such as CK-MB. Alternatively, in this particular patient subset with a lower peak CK-MB ratio, the prognostic significance of postoperative Q-waves may require a longer term follow-up to determine their importance.
The use of blood or crystalloid cardioplegia had no effect on survival. There was a favorable association between retrograde cardioplegia and a reduction of six-month mortality. Intermittent cross-clamping was associated with an increased six-month mortality rate. The duration of cardiopulmonary bypass and total surgical time did not significantly predict six-month mortality after adjustment for the peak CK-MB ratio and other co-variates.
The interpretation of our data may be complicated by several potential limitations. The patients enrolled in GUARDIAN comprised a clinical patient subset at increased risk of peri-procedural myocardial necrosis, and it is unclear whether the observed relationship can be extrapolated to patients who are at lower clinical risk at the time of the operation. Although baseline variables known to impact mortality were controlled for in the multivariate models, the analysis may not have included complete consideration of all of the important coronary morphologic characteristics on the angiogram or all of the procedural variables that may have had an impact on prognosis. The number of CK-MB samples obtained did not allow for calculation of the AUC for CK-MB release. Nevertheless, the increased mortality risk associated with elevated CK-MB enzymes postoperatively indicates that these patients most likely have had a postoperative MI and should be considered for drug therapy, such as beta-blockers and angiotensin-converting enzyme inhibitors, to decrease the postoperative risk of death (26,27). More research is needed in this area to determine whether cardioactive drugs that have been shown to be beneficial in medically treated post-infarction patients reduce mortality in postoperative patients with elevated CK-MB values. The highest cariporide dose investigated in the GUARDIAN trial reduced the mean CK-MB ratio elevation. This reduction was associated not only with a lowering of the primary composite end point of death or MI manifested early during the treatment phase, but also with the long-term secondary end point of events related to left ventricular dysfunction. Death rates, however, were unaffected at any time point (10). Cardiac troponin I measurements were available in only a small number of GUARDIAN patients (n = 176). The use of cardiac troponin measurements to predict outcome after CABG is an area of active research, although there are currently no data from large-scale clinical trials to correlate the magnitude of troponin release to medium- or long-term clinical outcomes (3,28–30). Our data apply to patients who have undergone cardiopulmonary bypass and cardioplegic arrest. The results have not been tested in patients who have off-pump surgery. Cardiac serum marker release after off-pump surgery is generally lower than that after on-pump surgery (31).
Our findings support the routine collection of CK-MB enzyme levels after clinically high-risk CABG, not only for diagnostic purposes, but also for prognostic guidance and treatment that may potentially reduce the high six-month mortality rates seen in patients with larger amounts of CK-MB enzyme release.
↵2 Drs. Bartels and Boyce served as consultants to Aventis Pharmaceuticals, Inc. Drs. Chaitman and Theroux served as consultants and received research grants from Aventis Pharmaceuticals, Inc. Ms. Keller and Dr. Jessel are employees of Aventis Pharmaceuticals, Inc.
↵1 A complete listing of the GUARD during Ischemia Against Necrosis (GUARDIAN) Investigators is found in Circulation 2000;102:3032.
- area under the curve
- coronary artery bypass graft surgery
- creatine kinase-myocardial band
- electrocardiogram or electrocardiographic
- GUARD during Ischemia Against Necrosis
- myocardial infarction
- percutaneous transluminal coronary angioplasty
- receiver operating characteristic
- upper limits of normal
- Received February 9, 2001.
- Revision received May 29, 2001.
- Accepted June 15, 2001.
- American College of Cardiology
- Califf R.,
- Abdelmeguid A.E.,
- Kuntz R.,
- et al.
- Hake U.,
- Iversen S.,
- Sadony V.,
- et al.
- TIMI II Investigators
- Alpert J.S.,
- Thygesen K.,
- Antman E.,
- et al.
- Theroux P.,
- Chaitman B.R.,
- Erhardt L.,
- et al.
- ↵Theroux P, Chaitman BR, Danchin N, et al., for the GUARD during Ischemia Against Necrosis (GUARDIAN) Investigators. Inhibition of the sodium–hydrogen exchanger with Cariporide to prevent myocardial infarction in high-risk ischemic situations: main results of the GUARDIAN trial. Circulation 2000;102:3032–8.
- Chaitman BR, Rosen AD, Williams DO, et al., for the BARI Investigators. Cardiac mortality and myocardial infarction in the Bypass Angioplasty Revascularization Investigation (BARI) randomized trial. Circulation 1997;96:2162–70.
- ↵Armitage P, Berry G. Statistical Methods in Medical Research. Cambridge, MA: Blackwell Scientific Publications, 1994:397–99, 453–60.
- Kalbfleisch J.D.,
- Prentice R.L.
- Brooks MM, Jones RH, Bach RG, et al., for the BARI Investigators. Predictors of mortality from cardiac causes in the Bypass Angioplasty Revascularization Investigation (BARI) randomized trial and registry. Circulation 2000;101:2682–9.
- Force T.,
- Kemper A.J.,
- Bloomfield P.,
- et al.
- Mair P.,
- Mair J.,
- Seibt I.,
- Antretter H.,
- Balogh D.,
- Puschendorf B.
- Chaitman B.R.,
- Alderman E.L.,
- Sheffield L.T.,
- et al.
- Pfeffer M.A.
- Sadony V.,
- Korber M.,
- Albes G.,
- et al.
- Ascione R.,
- Lloyd C.T.,
- Gomes W.J.,
- Caputo M.,
- Bryan A.J.,
- Angelini G.D.