Author + information
- Received September 6, 2001
- Revision received January 30, 2002
- Accepted February 6, 2002
- Published online May 1, 2002.
- James L Januzzi Jr, MD*,
- Kent Lewandrowski, MD†,
- Thomas E MacGillivray, MD‡,
- John B Newell, AB‡,
- Sekar Kathiresan, MD*,
- Stephen J Servoss, MD* and
- Elizabeth Lee-Lewandrowski, PhD, MPH†,* ()
- ↵*Reprint requests and correspondence:
Dr. Elizabeth Lee-Lewandrowski, Clinical Chemistry Laboratories, Massachusetts General Hospital, Gray-Bigelow 5, 55 Fruit Street, Boston, Massachusetts 02114, USA.
Objectives The aim of this study was to assess the role of serum markers of myocardial necrosis after cardiac surgery.
Background The role of serum troponin T (TnT) and creatine kinase-MB (CK-MB) for the risk stratification of patients after cardiac surgery remains undefined.
Methods Serum levels of TnT and CK-MB were measured from 224 patients every 8 h after cardiac surgery. The results of serum cardiac marker testing were correlated with adverse events, including new myocardial infarction (MI), cardiogenic shock or death. Univariable analysis identified factors predictive of complications, while stepwise logistic regression identified independent predictors of postoperative complications.
Results Cardiac marker elevation was universal after cardiac surgery. At all time points measured, compared with those patients without complications, the TnT levels from patients with complications were more significantly elevated (all: p < 0.0005). In contrast, among identically timed specimens, the levels of CK-MB from complicated patients were less reliably discriminatory. Multivariable analysis suggested that a TnT level in the highest quintile (≥1.58 ng/ml) was the strongest predictor of complications, including death (post-op, odds ratio [OR] = 31.0, 95% confidence interval [CI] = 3.67 to 263.1, p = 0.002) or shock (post-op: OR = 18.9, 95% CI = 2.29 to 156.1, p = 0.006; 18 h to 24 h: OR = 30.7, 95% CI = 3.75 to 250.7, p = 0.001), as well as the composite end points of death/MI (18 h to 24 h: OR = 60.1, 95% CI = 7.34 to 492.1, p < 0.0005), shock/MI (post-op: OR = 23.3, 95% CI = 2.82 to 191.4, p = 0.003; 18 h to 24 h: OR = 37.8, 95% CI = 4.66 to 307.3, p = 0.001) or death/shock/MI (post-op: OR = 20.0, 95% CI = 2.81 to 142.0, p = 0.003; 18 h to 24 h: OR = 67.4, 95% CI = 6.96 to 652.3, p < 0.0005). In contrast, in the presence of TnT, the results of CK-MB measurement added no independent prognostic information.
Conclusions Troponin T is superior to CK-MB for the prediction of impending complications after cardiac surgical procedures.
Among patients undergoing most cardiac surgery procedures, a measurable degree of myocardial injury is expected. In most cases, this injury is nothing more than an epiphenomenon of the procedure, but, in certain circumstances, significant myocardial necrosis may result. This process of perioperative myocardial infarction (MI) may exist as a focal MI in a single coronary artery territory, which may be easily identifiable on a 12-lead electrocardiogram (ECG) or may present as diffuse myocardial injury, which is more difficult to clinically recognize, occasionally presenting as severe, unexplained postoperative instability.
The current standard biochemical method for the detection of perioperative myocardial necrosis is the serial assessment of MB isoenzyme of serum creatine kinase (CK-MB), a marker that has significant limitations for the perioperative evaluation of patients undergoing cardiac surgery due to its limited sensitivity and specificity in this setting (1). Because of the limitations of CK-MB after cardiac surgery, a more sensitive and specific biochemical marker for postoperative biochemical risk stratification would be desirable. The serum troponins (both I and T) are more sensitive and specific than CK-MB in the setting of acute coronary syndromes (2)and have been validated for postoperative risk stratification for noncardiac surgical procedures (3,4); however, their role in the evaluation of cardiac surgery patients has not been well defined. Accordingly, we systematically studied the diagnostic and prognostic role of postoperative assessment of serum troponin T (TnT), as compared with CK-MB after cardiac surgical procedures.
Patient population and clinical data acquisition
All study procedures were approved by the hospital Institutional Review Board. A total of 224 consecutive patients admitted to the cardiac surgical intensive care unit (ICU) at the Massachusetts General Hospital were enrolled. Patients were identified on admission to the ICU, and a study coordinator collected clinical variables in a prospective manner via chart review, blinded to the results of cardiac markers. Among the clinical factors collected included demographics, past medical history, prior medication use, cardiac catheterization results and presenting cardiac syndrome. Information regarding surgical procedures, including type of surgery, number of bypass grafts, bypass/ischemic times and intraoperative complications was noted. Postoperative data including ICU length of stay, the results of postoperative ECG and postoperative outcomes were then assessed, and the presence of complications was recorded. Patients were followed through the end of their hospitalization for outcomes.
Complications were prospectively defined as: 1) new electrocardiographic Q-wave MI, defined as new pathologic Q waves or new left bundle branch block in the absence of valve replacement procedures; 2) shock, defined as a sustained impairment in cardiac index (<1.5 l/min/m2) requiring any of the following: prolonged (≥24 h) or significant need (≥2 drugs) for vasopressor therapy, intraaortic balloon pump insertion or surgical re-exploration; and 3) death. End points were adjudicated to the end of hospitalization.
Biomarkers of myocardial necrosis
Serial samples of blood, approximately 5 ml each, were drawn on arrival to the surgical ICU (“post-op”), at 6 h to 8 h and 18 h to 24 h after surgery. The samples were assayed for CK-MB mass utilizing a commercially available assay (Roche Diagnostics, Indianapolis, Indiana). The conventional threshold of this assay for the diagnosis of myocardial necrosis at our institution is 6.9 ng/ml. Levels of serum TnT were measured utilizing the third generation Enzymun cTnT assay (Roche Diagnostics, Indianapolis, Indiana). The lower limit of detection of this TnT assay is 0.01 ng/ml, and the conventional threshold for the diagnosis of myocardial necrosis at our institution is 0.10 ng/ml. Both assays were performed on an Elecsys 1010 platform (Roche Diagnostics, Indianapolis, Indiana). The results of biomarkers of myocardial necrosis were obtained in a blinded fashion to the study investigators; however, the physicians caring for the patients were not blinded to the CK-MB results because this marker is routinely measured at our institution after cardiac surgery.
The mean levels of TnT and CK-MB among the study patients were assessed with respect to the presence or absence of pre-specified complications. The prediction of postoperative complications was then assessed with both clinical and biochemical strategies. Those patients with postoperative complications were compared with those with uncomplicated ICU courses, with respect to clinical factors including baseline demographics, past medical history, preoperative medications, presenting syndrome, surgical considerations, as well as the mean levels of postoperative markers of myocardial necrosis.
Mean cardiac marker levels were assessed for the group as a whole, as well as for those patients with or without complications. Comparisons of mean cardiac marker levels between patients with or without complications were made by log-transforming the data first, then performing multivariate analysis of variance (MANOVA) with post hoc Bonferroni-corrected pairwise comparisons. These tests were conducted with the SYSTAT 10 software (SPSS, Inc., Chicago, Illinois). In order to analyze the prognostic influence of an elevated TnT level, marker levels were log-transformed and divided into quintiles. Univariable analysis identified clinical and biochemical factors predictive of complications and was followed by a multivariable analysis utilizing stepwise logistic regression to identify independent predictors of postoperative complications. The logistic regression analysis used forward stepping with verification of goodness of fit, with the Hosmer-Lemeshow test. All candidate independent variables were started out of the logistic model, and each was entered (in order of largest to smallest test chi-square statistic) only while the maximum likelihood estimate of the corresponding regression parameter was significantly different from zero at p < 0.05. An independent variable was removed from the model only when its corresponding regression parameter was not significantly different from zero at p > 0.1. Logistic regression analysis was conducted with SYSTAT software (SPSS, Inc., Chicago, Illinois). Odds ratios (OR) were generated and expressed with 95% confidence intervals (CI). All p values are two-sided, with p values <0.05 considered significant.
Patient demographics and clinical factors, all patients
The baseline demographics of the study patients (including past medical history, extent of coronary artery disease, presenting cardiac syndrome and preoperative medication use) are detailed in Table 1and are comparable to studies of similar postoperative populations. The majority of study patients presented with acute coronary syndromes, including unstable angina pectoris (26.7%), non–Q-wave MI (14.2%) or ST-segment elevation MI (5.8%), while a significant percentage presented with congestive heart failure (30.8%). The balance of the patients consisted of those presenting with medically refractory stable angina pectoris (20.5%) or cardiac arrest (1.8%).
The surgical procedures performed during the study are detailed in Table 2. The majority of study patients underwent coronary artery revascularization either as a primary procedure (60.2%) or in combination with valve repair/replacement or aortic root replacement (15.9%). The balance of patients underwent various procedures without coronary revascularization, including valve replacement/repair surgeries (12.8%), aortic grafting procedures (with and without aortic valve replacement, 5.8%), pericardial stripping/pericardiectomy (1.8%), cardiac transplantation (1.3%), left ventricular assist device insertion (0.9%) and atrial septal defect closure (0.9%).
The mean ischemic time for the study patients was 97.1 ± 53.1 min (range: 0 to 313), with 14 patients (6.25%) undergoing off-pump coronary artery bypass graft (CABG). The mean number of CABG grafts implanted was 2.53 ± 1.79 (range: 0 to 8). Several different forms of myocardial protection strategies were employed with cold cardioplegia delivered anterograde and retrograde being the most common method, used in 35.6% of cases, followed by cold blood anterograde (17.5%), warm blood anterograde/retrograde (15.8%), cold fibrillatory arrest (14.3%), warm blood anterograde (7.2%), off-pump surgery (6.3%) and cold crystalloid (3.3%). Among the surgeries employing aortic root replacement, 11 patients (4.9%) underwent deep hypothermic circulatory arrest, in addition to receiving cardioplegia.
Six patients (2.7%) developed new ECG evidence for MI, and 16 (7.1%) developed postoperative cardiogenic shock. There were nine deaths (4.0%), of which eight (3.6%) were of cardiac cause and occurred in the ICU; one patient (0.4%) died of pneumonia later in their hospitalization.
Cardiac marker levels
The mean cardiac marker levels of all patients in the study are detailed in Table 3. The mean immediate post-op, 6 h to 12 h and 18 h to 24 h levels of TnT were 1.87 ± 3.99 ng/ml (range: <0.01 to 50.04 ng/ml), 2.14 ± 3.44 ng/ml (range: <0.01 to 37.40 ng/ml) and 1.58 ± 3.52 ng/ml (<0.01 to 35.58 ng/ml), while the results for identically timed specimens of CK-MB were 66.58 ± 67.26 ng/ml, 67.31 ± 71.30 ng/ml and 34.80 ± 42.45 ng/ml, respectively.
Compared with patients without complications, patients with a complicated ICU course (n = 21) were more likely to have higher levels of both TnT and CK-MB at all time points examined (Table 3). Compared with uncomplicated patients, the TnT and CK-MB levels among complicated patients postoperatively, at 6 h to 12 h and 18 h to 24 h, were significantly elevated. For TnT, the levels were 8.88 ± 10.54 ng/ml vs. 1.13 ± 0.86 ng/ml postoperatively, 8.59 ± 8.42 ng/ml vs. 1.46 ± 1.05 ng/ml at 6 h to 12 h and 8.43 ± 9.10 ng/ml vs. 0.89 ± 0.75 ng/ml at 18 h to 24 h (all: p < 0.0005). The results for identically timed specimens of CK-MB were 152.83 ± 145.01 ng/ml vs. 57.48 ± 44.80 ng/ml, 164.55 ± 156.19 ng/ml vs. 57.15 ± 45.62 ng/ml and 75.02 ± 71.16 ng/ml vs. 30.60 ± 36.02 ng/ml (all: p < 0.0005).
The cardiac marker levels of those patients with individual complications are detailed in Table 4. Those patients with new ECG evidence for MI (n = 6) were more likely to have elevations of both TnT (post-op: 8.41 ± 6.46 ng/ml vs. 1.13 ± 0.86 ng/ml; 6 h to 12 h: 5.96 ± 3.17 ng/ml vs. 1.46 ± 1.05 ng/ml; 18 h to 24 h: 6.00 ± 3.90 ng/ml vs. 0.89 ± 0.75 ng/ml; all: p < 0.0005), as well as CK-MB, with the exception of the 6 h to 12 h CK-MB specimen, which was not significantly different (post-op: 202.23 ± 212.89 ng/ml vs. 57.48 ± 44.80 ng/ml, p < 0.0005; 18 h to 24 h: 79.90 ± 54.40 ng/ml vs. 30.60 ± 36.02 ng/ml, p < 0.05). Those patients with shock (n = 16) were more likely to have statistically significantly higher TnT levels than those without shock, at all three time points examined (post-op: 8.59 ± 11.61 ng/ml vs. 1.13 ± 0.86 ng/ml; 6 h to 12 h: 9.03 ± 9.52 ng/ml vs. 1.46 ± 1.05 ng/ml; 18 h to 24 h: 10.13 ± 9.98 ng/ml vs. 0.89 ± 0.75 ng/ml; all: p < 0.0005). Similarly, the levels of identically timed CK-MB specimens from the patients with shock were significantly different from those without (post-op: 123.20 ± 123.19 ng/ml vs. 57.48 ± 44.80 ng/ml, p < 0.05; 6 h to 12 h: 141.29 ± 134.22 ng/ml vs. 57.15 ± 45.62 ng/ml, p = 0.001; 18 h to 24 h: 76.20 ± 76.53 ng/ml vs. 30.60 ± 36.02 ng/ml, p < 0.05). Finally, in-hospital death (n = 9) was associated with significantly higher levels of TnT at all time points analyzed (post-op: 9.71 ± 15.52 ng/ml vs. 1.13 ± 0.86 ng/ml; 6 h to 12 h: 10.79 ± 12.42 ng/ml vs. 1.46 ± 1.05 ng/ml; 18 h to 24 h: 12.73 ± 13.11 ng/ml vs. 0.89 ± 0.75 ng/ml; all: p < 0.0005) compared with those patients who survived hospitalization. In contrast, there was no significant difference in CK-MB levels between patients who died compared with survivors.
Predictors of complications—multivariable analysis
Among each of the multivariable analyses performed for the prediction of complications (which included both clinical variables and biomarker results), a TnT level in the highest log quintile (≥1.58 ng/ml) immediately postoperatively or at 18 h to 24 h was the strongest predictor of individual postoperative complications, including death (18 h to 24 h: OR = 31.0, 95% CI = 3.67 to 263.1, p = 0.002) or shock (post-op: OR = 18.9, 95% CI = 2.29 to 156.1, p = 0.006; 18 h to 24 h: OR = 30.7, 95% CI = 3.75 to 250.7, p = 0.001), as well as the composite end points of death/MI (18 h to 24 h: OR = 60.1, 95% CI = 7.34 to 492.1, p < 0.0005), shock/MI (post-op: OR = 23.3, 95% CI = 2.82 to 191.4, p = 0.003; 18 h to 24 h: OR = 37.8, 95% CI = 4.66 to 307.3, p = 0.001) or death/shock/MI (post-op: OR = 20.0, 95% CI = 2.81 to 142.0, p = 0.003; 18 h to 24 h: OR = 67.4, 95% CI = 6.96 to 652.3, p < 0.0005).
Of note, the results of the 6 h to 12 h TnT specimen did not add independent prognostic information to the results of the postoperative or 18 h to 24 h TnT assays. In contrast to the TnT results, none of the identically timed CK-MB levels added significant independent prognostic information in the presence of TnT results.
Of the patients with available TnT data preoperatively, only 26 had a level ≥0.10 ng/ml. No clear association between elevated preoperative TnT and adverse postoperative outcomes was noted. Furthermore, excluding these patients from analysis of the group as a whole did not alter the significantly prognostic significance of elevated postoperative TnT.
Because of superior sensitivity and specificity, the serum troponins have replaced CK-MB as the gold standard for the biochemical diagnosis of myocardial injury (2)in several clinical situations, including the risk stratification of patients with acute coronary syndromes (5,6), as well as those undergoing noncardiac surgery (3,4). The role of the troponins for the evaluation of patients after cardiac surgery remains less clear, however. Several studies have evaluated troponin testing for this purpose (7–16)but were limited by small numbers of patients, low event rates and usually excluded most patients with procedures other than CABG. In addition, studies occasionally returned conflicting results with respect to the value of troponins in the cardiac surgery setting (17). We, therefore, undertook a prospective study evaluating the use of TnT assessment for the postoperative evaluation of a larger group of cardiac surgery patients. In addition, we compared the diagnostic and prognostic value of TnT to CK-MB, the current gold standard for biochemical postcardiac surgery risk stratification.
We found that significant elevations of TnT are nearly ubiquitous after cardiac surgical procedures. Despite this fact, the mean levels of TnT were significantly higher among those patients who ultimately developed a complicated postoperative course. In contrast, although elevated to a similar magnitude as TnT in many complicated cases, the results of postoperative CK-MB levels were not as discriminatory of impending severe complications such as death, demonstrating very wide standard deviation. Finally, we found in multivariable analysis that a TnT level ≥1.58 ng/ml in the immediate postoperative and 18 h to 24 h TnT specimens offered the strongest independent prognostic information regarding postoperative complications, while CK-MB levels added no independent prognostic information in the presence of TnT results.
Troponin T levels after cardiac surgery
The TnT levels that we demonstrated among most of the study population were frequently more than 10× above the upper limit of normal, even in the absence of evident complications. These results are consistent with the findings from earlier studies, where similar ranges of TnT levels among uncomplicated patient populations were noted (7,18,19). Furthermore, although the TnT cut-off of 1.58 ng/ml, which we identified as predictive of complications in our study, is more than 15× above the upper limits of normal for this assay, it was still possible to easily discern those patients with complications from those patients without. The magnitude of TnT elevations in our study are similar to the findings reported for patients with and without complications in studies of troponin I by Greenson, et al. (16)as well as Immer et al. (13,14)among adult and pediatric cardiac surgery populations, respectively. It is of note that TnT levels were not universally outside of the conventional reference range, with some patients demonstrating only slightly elevated levels (such as those undergoing off-pump CABG) and even some patients with levels of TnT less than the detectable range (such as those with pericardial stripping). Therefore, further characterization of the “expected” release of TnT after various types of surgery is necessary as the use of this marker increases among postcardiac surgery patients. In addition, as TnT is thought to be entirely cardiac-specific, further defining the characteristics of TnT release in this patient population may offer improved opportunities to evaluate novel strategies for myocardial preservation/protection.
Cardiac markers, postoperative clinical evaluation and outcomes
While a significant number of patients developed elevated TnT levels postoperatively in conjunction with symptoms/signs of myocardial dysfunction such as cardiogenic shock and death, very few developed ECG evidence for new pathologic Q waves. Although reported to occur in as few as 1.3% and as many as 25% of patients undergoing cardiac surgery (20), electrocardiographic Q-wave MI is only one manifestation of postoperative myocardial necrosis. Indeed, other states, including non–Q- wave MI as well as diffuse myocardial injury, may occur and may be much more difficult to clinically recognize. Our data would suggest that the latter two processes are more common than previously suspected and further confirm the concept that ECG is insensitive for most forms of postcardiac surgery myocardial injury states (21). Our data suggest that TnT measurement in the postoperative cardiac surgery patient is particularly useful for the diagnostic evaluation of unexplained postoperative patient instability and may be superior to clinical exam or diagnostic testing such as ECG for this purpose. This is consistent with the results reported by Jacquet et al. (22)who noted that troponin I assessment was reliable for the identification of excessive myocardial necrosis in the setting of post-CABG instability, particularly in the setting of a nondiagnostic ECG. However, in contrast with the study by Jacquet and colleagues, our data further suggests that measurement of serum TnT not only allows for the identification of those patients with significant myocardial necrosis from those without but also provided the strongest independent prediction of severe postoperative complications, including death.
While our study is one of the largest systematic assessments of serum TnT for the evaluation of postcardiac surgery patients to date, it has limitations. First, the relatively small number of outcomes limits power for estimation of the prognostic role for TnT, especially for individual outcomes. Furthermore, another potential source of weakness is that our study enrolled patients undergoing a wide variety of surgical procedures, with various forms of myocardial protection and postoperative management strategies. This variation in cardioplegia and surgical strategies are reflective of general cardiac surgical practice, however, and the prognostic role of TnT remained significant, demonstrating the flexibility of this marker for the assessment of patients undergoing various types of surgery. Additionally, not all patients in the study had their preoperative TnT level assessed; thus, we could not systematically examine the effects of prior elevation of serum TnT on the postoperative marker levels or postoperative complications. However, among those patients with elevated preoperative TnT levels (such as those entering the study with an acute coronary syndrome), there was no apparent incremental risk for complications above that of postoperative TnT above the threshold of 1.58 ng/ml. Not surprisingly, therefore, excluding those patients with elevated preoperative serum TnT from the group as a whole did not change the strong prognostic value of elevated TnT in the postoperative setting. Finally, as the follow-up in our study extends only to the end of the hospitalization, little can be generalized about the long-term prognostic role of TnT in this patient population. We are currently in the process of analyzing the longer-term prognostic implications of serum TnT elevation after cardiac surgery in this patient population.
In conclusion, we have demonstrated that after cardiac surgical procedures, assessment of serum TnT immediately postoperatively as well as at 18 h to 24 h after surgery provided a widely available, simple, rapid, inexpensive and accurate method for identifying patients with significant perioperative myocardial necrosis. Furthermore, in our study, assessment of serum TnT added powerful prognostic information regarding the impending development of severe postoperative complications, including death, as well as adding confirmatory information in patients with unexplained postoperative hemodynamic instability. In contrast, CK-MB levels were less reliable for the discernment of postoperative complications, and, in the presence of TnT, results added no independent prognostic information in this postoperative group. Our data suggest that TnT is superior to CK-MB for the postoperative evaluation of patients after cardiac surgical procedures and, additionally, may offer new opportunities to evaluate perioperative myocardial protection strategies. Based on these results, the replacement of CK-MB with serum troponin testing for the postoperative evaluation of the cardiac surgical patient should be strongly considered.
The authors particularly thank Steven McDermott, RN, for his outstanding contributions to the study, Christopher J. O’Donnell, MD, for his insightful comments on the manuscript and the entire nursing staff of the Massachusetts General Hospital, Blake 8 Cardiac Surgical Intensive Care Unit, without whose diligent efforts this study would not have been possible.
☆ Supported, in part, by an unrestricted grant from Roche Diagnostics, Indianapolis, Indiana.
- coronary artery bypass graft
- confidence interval
- creatine kinase-MB
- intensive care unit
- myocardial infarction
- odds ratio
- troponin T
- Received September 6, 2001.
- Revision received January 30, 2002.
- Accepted February 6, 2002.
- American College of Cardiology Foundation
- The Joint European Society of Cardiology/American College of Cardiology Committee
- Hake U,
- Schmid F.X,
- Iversen S,
- et al.
- Immer F.F,
- Stocker F,
- Seiler A.M,
- et al.
- Brasch A.V,
- Khan S.S,
- Denton T.A,
- et al.
- Svedjeholm R,
- Dahlin L.G,
- Lundberg C,
- et al.
- Jacquet L,
- Noirhomme P,
- El Khoury G,
- et al.