| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Anesthesia and Intensive Care Medicine, University Hospital Basel, Basel, Switzerland
b Basel Institute of Clinical Epidemiology, University Hospital Basel, Basel, Switzerland
c Division of Cardiothoracic Surgery, University Hospital Basel, Basel, Switzerland
Accepted for publication July 31, 2009.
* Address correspondence to Dr Lurati Buse, Department of Anesthesia and Intensive Care Medicine, University Hospital Basel, Spitalstrasse, Basel, CH-4031, Switzerland (Email: luratig{at}uhbs.ch).
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Methods: This cohort study included consecutive patients undergoing on-pump cardiac surgery between January 2005 and December 2006. We evaluated postoperative troponin T (TNT) on days 1 and 2 and the EuroSCORE as predictor variables. The primary composite endpoint was all-cause mortality or any major adverse cardiac event (MACE) at 12 months. Logistic regression was used to study the prognostic effect of TNT in a univariate analysis and after adjustment for EuroSCORE. The area under the receiver-operator curve (AUC) was calculated to report the discriminatory performance of the models.
Results: Seven hundred forty-one patients were available for analysis. Within 12 months after surgery, 92 (12.4%) patients had a MACE, 48 (6.5%) of whom died. A multivariate model of continuous TNT and the continuous logistic EuroSCORE showed a significant independent association between TNT and the composite endpoint (odds ratio [OR] 1.03, 95% confidence interval [CI] 1.02 to 1.04 per 0.1 µg/L increase in TNT). The AUC for the prediction of the composite endpoint of the model combining TNT and the EuroSCORE was 0.72; when based on EuroSCORE alone it was 0.64 (p < 0.0001).
Conclusions: Postoperative TNT increase (per 0.1 µg/L) is a strong independent predictor of 12-month outcome after on-pump cardiac surgery. Updating the preoperative EuroSCORE risk with postoperative TNT allows for better prediction of 12-month MACE and all-cause mortality.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Laboratory variables that might integrate perioperative cardiac events into risk stratification models are troponin I and troponin T. Troponin T (TNT) has an established prognostic value in nonsurgical populations [8] and several authors have reported on its independent prognostic value for short-term outcome in cardiac surgery [9–11]. However, its value for midterm outcome in cardiac surgical patients is insufficiently known.
The EuroSCORE is a widely used preoperative risk assessment tool in clinical practice. It considers patient-related risk factors such as age and gender, comorbidities, cardiac factors such as ejection fraction or recent myocardial infarction, and the type of surgery [2].
The aim of this cohort study was to test the hypothesis that postoperative TNT release is associated with 12-month outcome in adults undergoing on-pump cardiac surgery. A secondary aim of the study was to investigate the increase in prognostic accuracy by supplementing the EuroSCORE [2, 3] with TNT; ie, by adding a postoperative variable to an established preoperative risk score.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Perioperative Management
Patients were orally premedicated with midazolam (7.5 mg) or bromazepam (1.5 to 3 mg).
Anesthesia was induced by thiopental (3 to 4 mg/kg) or etomidate (0.2 to 0.3 mg/kg) and fentanyl (2 to 6 µg/kg) and it was maintained by isoflurane and by midazolam-fentanyl-infusion during cardiopulmonary bypass (CPB), respectively. During CPB isoflurane administration was continued. Neuromuscular blockade was induced and maintained by atracurium.
Heparin (350 U/kg) was administered with a target activated clotting time greater than 480 seconds. At the same time as the initial heparin dose, tranexamic acid (30 mg/kg) was given. Cardiopulmonary bypass priming volume amounted to 2,000 mL. Myocardial protection was achieved by intermittent anterograde blood or crystalloid cardioplegia. The body temperature was maintained at greater than 32°C. The CPB weaning was managed with catecholamines at the discretion of the attending anesthesiologist. Heparin reversion was achieved by protamine. Blood was transfused if hemoglobin fell below 80 g/L. Patients were admitted postoperatively to the intensive care unit for 48 hours.
Baseline Preoperative Variables
Preoperative baseline data on patients included cardiac history, New York Heart Association functional state, cardiac medication, and comorbidities, in addition to the data profiles required to calculate the EuroSCORE. The logistic EuroSCORE was used for all analyses [3]. Hematologic, basic renal, hepatic, and metabolic laboratory results were collected.
Perioperative Surgical and Clinical Variables
We collected detailed data on perioperative patient treatment, surgical techniques, and postoperative course. Operative data included operating time, blood loss, cardiopulmonary and cross-clamp time, number of arterial and venous grafts, number of distal anastomoses, and the identity of the surgeon.
Cardiac Marker Analyses
Troponin T (Elecsys; Roche Diagnostics, F. Hoffmann-La Roche Ltd, Basel, Switzerland) and MB isoenzyme of creatine kinase (CK-MB) concentrations (ECLIA, Roche Diagnostics, F. Hoffmann-La Roche Ltd) were measured routinely at 6 am on the first and second postoperative day in all patients. All analyses were performed by the local hospital laboratory, which is subjected to practice-certified regulatory quality controls.
Clinical Endpoint Definitions and Endpoint Ascertainment
The primary endpoint was defined as the composite of death from any cause or the occurrence of nonfatal major adverse cardiac events (MACE), defined as myocardial infarction, need for subsequent surgical or percutaneous coronary intervention, and congestive heart failure requiring hospitalization within one year after surgery. Secondary analyses were restricted to the endpoint all-cause mortality.
Data on events were prospectively obtained by a questionnaire, mailed to all patients one year after surgery. Patients or next of kin were asked to fill in a simple questionnaire on the occurrence and date of any hospitalization within one year after surgery or the patient's vital status, respectively. If the questionnaire was not returned, we contacted the patient by telephone for a semistructured interview. The interviews were conducted by a trained study nurse, blinded to all patient data. If the questionnaire or the interview indicated that the patient was admitted to a hospital for any reason, we contacted the hospital or the patient's general practitioner to obtain the discharge letters. Based on this information, the occurrence of an endpoint according to the prespecified definition was determined by two independent assessors blinded to all patient data. If direct contact with a patient could not be established, we asked the patient's general practitioner to provide the required health status information.
Statistical Analysis
Descriptive statistics
Baseline data of the participants are presented as mean ± SD or as median with interquartile ranges (IRQ, 25th and 75th percentiles) as appropriate. Categoric data are presented as absolute numbers and percentages. Baseline differences between patients with events compared with event-free patients were calculated with the Student t or Pearson 
2 tests. We assessed calibration of the logistic EuroSCORE in our population and plotted the predicted mortality estimated by the logistic EuroSCORE against the observed mortality in quartiles of the logistic EuroSCORE.
Univariable analysis
The higher of the two available TNT values was used for analyses. A first univariable logistic regression model was calculated by fitting the TNT values as a continuous variable. A second logistic regression model was calculated by fitting the TNT as a dichotomous (nonelevated versus elevated) variable. The cutoff value was defined by receiver operating characteristic (ROC) statistics with a 1:1 weighting of sensitivity and specificity representing the TNT concentration that achieved the best accuracy (calculated as 0.5 x [sensitivity + specificity]) for the prediction of the primary endpoint. These models provided unadjusted odds ratios (ORs). In addition, the second model also provided positive and negative likelihood ratio for the outcome measures. As a prespecified subgroup analysis, we tested for heterogeneity between surgery with versus without coronary intervention using the Tarone test for heterogeneity.
Multivariate analysis
We predefined the EuroSCORE as an adjustment variable in a multivariate logistic regression to assess the independent prognostic effect of TNT on the composite endpoint. We fitted a model including TNT and the logistic EuroSCORE as continuous variables in the logistic regression. The model including the EuroSCORE alone and the model extended with TNT were compared by the likelihood-ratio test.
Another model using the logistic EuroSCORE in quartiles and TNT as dichotomous variable was also fitted and tested for interaction by the likelihood-ratio test. To address the performance of the prognostic models, we expressed discrimination of the model including the EuroSCORE only, and of the model extended with TNT, as the area under the ROC curve (AUC).
Sample Size
For the 24-month enrollment period, we anticipated a sample size of 800 patients to reliably study the effect of TNT on the predefined endpoints. Estimating a 12-month event rate of 10% (all-cause mortality of 6% plus 4% MACE), we expected 80 events, allowing for a statistically robust multivariate model based on up to eight variables [12].
We considered a two-sided p value less than 0.05 as statistically significant. Models were calculated by Stata for Windows (release 9.2; Stata Corp, College Station, TX).
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Within the 12-month postoperative follow-up, 92 (12.4%) out of 741 patients experienced the primary endpoint. Forty-eight of these 92 events were deaths (12-month mortality 6.5%), 18 (19.5%) acute congestive heart failures requiring hospitalization, 8 (8.7%) myocardial infarctions, and 18 (19.5%) reinterventions (including 14-percutaneous transluminal coronary angioplasty). Compared with patients with uneventful follow-up, those who suffered an event had higher EuroSCORE, and higher TNT and CK-MB values. Details are provided in Table 1.
|
|
Multivariate Analysis: Independent Value of Troponin T
A multivariate model of continuous TNT and the continuous logistic EuroSCORE showed a significant independent association between TNT and the composite endpoint, and between TNT and 12-month all-cause mortality (Table 2). We proceeded to incorporate the logistic EuroSCORE into the model only after assessment of its calibration to the 12-month mortality in our cohort.
|
2 = 1.67, degrees of freedom (df) 3, p = 0.64).
The AUC for the ROC curve derived from the model with TNT and logistic EuroSCORE as continuous covariates was larger than the AUC from the model with logistic EuroSCORE as only covariate (Table 2). The likelihood-ratio test of the TNT extended model showed a better fit compared with the univariate EuroSCORE model (likelihood-ratio test 2 = 68.62, df 2, p < 0.0001 for the composite endpoint and likelihood-ratio test 2 = 75.21, df 2, p < 0.0001 for 12-month all-cause mortality; Table 2).
Subgroup Analysis
We did not find evidence for heterogeneity between procedures with versus without coronary artery surgery (unadjusted OR of TNT >0.9 µg/L 2.73 [1.59 to 4.69] vs OR 2.48 [1.01 to 6.11], respectively; p = 0.859).
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Risk stratification in cardiac surgery has so far focused on preoperative and intraoperative patient risk factors [1]. The EuroSCORE, for example, summarizes prognostic factors such as age and gender, comorbidities (eg, chronic obstructive lung disease, neurologic impairment), cardiac factors (eg, ejection fraction, recent myocardial infarction, pulmonary artery hypertension), and the type of surgery [2]. This scoring tool was shown to predict short-term [13] and long-term [14] mortality with a discriminative performance as measured by the area under the ROC curve, between 0.76 [2] and 0.86 [15] for short-term and 0.72 [14] for long-term mortality.
Patient outcome is not determined by the patient's preoperative condition alone, it also depends on perioperative events such as perioperative myocardial ischemia, which is associated with increased morbidity and mortality after cardiac surgery [4–7]. Perioperative ischemia is, however, difficult to detect because of the low diagnostic accuracy of electrocardiography [16, 17]. In addition, the association between ischemic electrocardiogram abnormalities and outcome measures is weak [18, 19].
Given the difficulty in detecting perioperative ischemia, troponin has been developed as an alternative prognostic marker in cardiac surgical patients. Evidence of a concentration-response relationship with troponin I release was presented by Adabag and colleagues [9] for short-term mortality (OR 1.3, [95% CI 1.1 to 1.5] per 0.05 µg/L increase in troponin I levels in valvular patients and 1.4 [95% CI 1.2 to 1.6] in CABG patients). Nesher and colleagues [10] reported a concentration-dependent association of TNT and short-term MACE and reported a strong association for a cutoff concentration of 0.8 µg/L for TNT (univariate OR 4.8 [95% CI 1.7 to 13.1] for short-term mortality). Interestingly, we obtained a TNT cutoff level at 0.9 µg/L (postoperative day 1) by ROC-curve analysis of our data, which is very near to the cutoff concentration (0.8 µg/L) Nesher and colleagues calculated from their data. In a large study, Croal and colleagues [20] showed an OR of 1.10 (95% CI 1.03 to 1.18) per 10 µg/L troponin I increase for one-year mortality. Troponin I data, however, cannot be easily extrapolated to other settings due to the different analysis methods available on the market.
The strength of association between troponin release and outcome in other clinical settings, such as acute coronary syndromes (ACS) [21] or heart failure [22, 23] was comparable with the association found in patients after cardiac surgery, reported here. Thus, the association between loss of myocardial cells and mortality lies in a similar range independent of the trigger for myocardial damage.
Troponin is not expected to be detectable in patients with no previous heart surgery. Accordingly, any measurable troponin release can be associated with adverse outcome and cutoff values for adverse outcome neared the upper reference limits for the troponin tests used [21–27]. In contrast, in patients undergoing cardiac surgery, surgical manipulation, and ischemia-reperfusion damage are expected to lead to an inevitable structural damage, associated with a certain degree of troponin release. Actually, the cutoff levels found in cardiac surgical populations, both in the present and in previous studies, were several times higher than the upper reference limit of the test kits [10, 28, 29]. These findings indicate that only troponin concentrations exceeding a certain level reflect a more than inevitable myocardial damage, which is associated with adverse outcome after cardiac surgery.
Our data show an important gain in discriminatory performance for both the composite endpoint and all-cause mortality at 12-months, after extending the EuroSCORE by the additional consideration of maximum postoperative TNT concentrations. The established risk assessment scores (eg, the EuroSCORE) use preoperative data and they allow for comparison of performance of different centers. Our data show that adding TNT to these risk scores allows for improved risk prediction for long-term outcome. Identification of patients at higher risk might be particularly important for those considered to be at low risk (ie, low EuroSCOREs) but demonstrating increased troponin levels postoperatively. Such identification might become the first step for intensifying treatment and improving outcome of a subpopulation of patients at increased perioperative risk. However, the feasibility and clinical value of such a strategy needs to be tested in future trials.
The ability to generalize based on the results of this study is supported first by its broad inclusion criteria covering virtually all patients undergoing on-pump cardiac surgery at our hospital. Second, findings based on TNT can be more easily adapted to other hospitals than those based on troponin I because there is only one commercially available method for TNT analysis. Third, blood sample collection at a fixed time (6 am) on postoperative day 1 and 2 contributed to a closer similarity to clinical practice than blood sample collection after a fixed time span postoperatively. Fourth, we adjusted the association of TNT concentration for the EuroSCORE, a validated [1, 2] and widely used preoperative scoring tool in cardiac surgery. Finally, the mortality rate after on-pump cardiac surgery in our cohort lies within the reported range [11, 20, 28, 29].
Limitations
We are aware of some limitations of our study. First, we assessed TNT concentration in all patients at 6 am on postoperative day 1 and 2, independent of the time of surgery. Therefore, the samples of postoperative day 1 were collected between 12 and 20 hours after surgery and the samples of postoperative day 2 between 36 and 44 hours after surgery. Blood sample collection at a fixed point of the day, however, conforms to clinical routine rather than to sample collection at a defined time after surgery.
A further disadvantage of our consistent timing of blood sample collection is the fact that TNT was not measured in the patients dying in tabula or in the patients dying shortly after surgery (deaths before 6 am on the first postoperative morning). Our results are, thus, conditional on survival until at least the first postoperative morning at 6 am.
Second, baseline data relied on data from hospitalization charts, although the rate of missing values was very low (0.4% missing TNT values). Third, follow-up assessment based on mailed questionnaires was subject to nonresponse bias. However, telephone interviews with patients and (or) their general practitioners allowed us to achieve a follow-up completeness of 98.3%. Our follow-up assessment may have also been biased by misclassification of outcomes, but the probability of forgetting a hospitalization seemed remote. Moreover, misclassification bias could be expected to be nondifferential as the patients were not aware of their exposure status, ie, TNT readings.
Further, we restricted logistic regression to postoperative TNT concentration and EuroSCORE as prespecified. Therefore, we cannot discount that the inclusion of variables not summarized in our model by the EuroSCORE or a different weighting of the EuroSCORE variables might have influenced the strength of the associations that we described. However, our aim was to estimate the added effect of postoperative TNT release and not the derivation of a new clinical prediction rule.
Finally, we focused on postoperative TNT concentrations only. Thielmann and colleagues [17] showed a prognostic value of elevated preoperative troponin I concentrations for short-term mortality. However, the high baseline risk of patients presenting a recent ACS and (or) emergent surgery is already considered by preoperative risk stratification (EuroSCORE) [2].
Postoperative TNT increase is independently and strongly associated with one-year outcome after on-pump cardiac surgery. The extension of the EuroSCORE, a preoperative risk stratification tool, with postoperative TNT concentrations allowed for a better prediction of 12-month all-cause mortality and the occurrence of MACE after on-pump cardiac surgery.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Related Article
This article has been cited by other articles:
|
|
 |
|
  S. Siregar, R. H. H. Groenwold, F. de Heer, M. L. Bots, Y. van der Graaf, and L. A. van Herwerden Performance of the original EuroSCORE Eur J Cardiothorac Surg, January 26, 2012; (2012) ezr285v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. K. Singh and S. E. Fremes Invited Commentary Ann. Thorac. Surg., December 1, 2009; 88(6): 1812 - 1813. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
