Abstract
BACKGROUND: The clinical relevance of slightly increased circulating troponin concentrations in patients with stable coronary heart disease (CHD) several weeks after an acute event or CABG has not been fully evaluated.
METHODS: Baseline plasma concentrations of troponin T were measured with a high-sensitivity assay (hs-cTnT) (Roche Elecsys) in a cohort of 1050 CHD patients from 30 to 70 years of age. The prognostic value of hs-cTnT on a combined cardiovascular disease (CVD) end point after adjustment for covariates was determined with Cox proportional hazards modeling.
RESULTS: The median hs-cTnT concentration was 10.9 ng/L (interquartile range, 5.1–18.9 ng/L). Increased hs-cTnT concentrations were associated with an older age, history of hypertension and diabetes, more advanced coronary artery disease, and other CHD risk factors. Furthermore, hs-cTnT concentration was strongly correlated with N-terminal pro–B-type natriuretic peptide (NT-proBNP) and cystatin C (ρ = 0.61, and ρ = 0.32, respectively; both P values <0.0001). During a median follow-up of 8.1 years, 150 patients (14.3%) experienced a secondary CVD event. In a multivariate model, hs-cTnT was associated with a hazard ratio (HR) for secondary events of 2.83 (95% CI, 1.68–4.79) when the extreme quartiles were compared. Further adjustment for cystatin C, NT-proBNP, and C-reactive protein attenuated this association only slightly (HR, 2.27; 95% CI, 1.31–3.95); P for trend < 0.002). ROC curve analysis of a clinical model that added hs-cTnT to a baseline model showed nonsignificant improvement in the area under the curve (0.69 vs 0.67), whereas the net reclassification improvement was 17.2% (P = 0.029).
CONCLUSIONS: Slightly increased hs-cTnT concentrations in stable CHD patients are associated with several cardiovascular disorders and predict long-term CVD events.
Cardiac troponins, sensitive markers of cardiomyocyte necrosis, are released into the blood stream after cell injury. Troponins have been firmly established for >10 years as diagnostic biomarkers in the setting of the acute coronary syndrome (ACS)6 (1, 2), and their concentrations have been clearly associated with prognosis (3). In accord with the clinical value of troponin measurements, the statement of the universal definition of myocardial infarction (MI) (4) has classified different MI types by using troponins as the preferred markers of necrosis, and a consensus report regarding the use of cardiac troponins in acute cardiac care has been published (5). More recently, so called high-sensitivity troponin T and I assays have been developed that can detect concentrations that are 10-fold lower than with currently available fourth-generation assays. For cardiac troponin T as measured with a high-sensitivity assay (hs-cTnT), the detection limit is 3 ng/L, the 99th percentile of the hs-cTnT distribution is 14 ng/L in the general population (6), and >50% of asymptomatic individuals should have a detectable value (7). Four recent reports (8,–,11) have highlighted the improved diagnostic accuracy of such high-sensitivity assays in the diagnosis of ACS, for both high-risk and low-risk individuals.
The residual risk in patients after manifest coronary heart disease (CHD) is still high despite optimal contemporary treatment, including acute percutaneous intervention and polypharmacotherapy. Thus, improved risk stratification is crucial for identifying patients who would be candidates for more-aggressive therapy and potential new therapeutic modalities. Such efforts have been undertaken via the introduction of various scores, such as the TIMI (Thrombolysis in Myocardial Infarction) score or the GRACE (Global Registry of Acute Coronary Events) score, that are based on measuring clinical and biochemical variables during ACS (12,–,14); however, these scores are not well established for stable patients, e.g., after coronary artery bypass grafting (CABG). In addition, biomarkers such as natriuretic peptides (15) have been investigated in this setting. Given the strong prognostic power of increased circulating concentrations of troponins in ACS and the ability of the new high-sensitivity assays to reliably detect low troponin concentrations, it is of interest to evaluate patients with stable CHD who are known to be at high residual risk and to assess whether troponin measurements obtained with these new assays can improve risk stratification. Omland et al. (16) studied 3593 patients with stable CHD and a preserved left ventricular function from the PEACE (Prevention of Events with Angiotensin-Converting Enzyme Inhibition) trial and followed them for a median of 5.2 years. They found a strong and graded increase in the cumulative incidence of death and heart failure but, surprisingly, no association with fatal or nonfatal MI. Further evidence for troponin measured with a high-sensitivity assay (in this case hs-TnI) to predict cardiovascular death in stable high-risk patients comes from another trial (n = 2572), the Heart Outcomes Prevention Evaluation (HOPE) study (17). In contrast with the findings from the PEACE trial, however, increased hs-TnI concentrations in the HOPE trial were also associated with future MI during a 4.5-year follow-up. This finding even held in a model that already included N-terminal pro–B-type natriuretic peptide (NT-proBNP). The risk estimates for cardiovascular death in the 2 studies were essentially the same.
Still, several aspects regarding the determinants of hs-cTnT in patients with manifest CHD and the relationship of marker concentrations with long-term outcomes need to be elucidated in more detail, especially in unselected patients or in routine medical care. Thus, we sought to assess determinants of hs-cTnT in a large, well-characterized cohort of stable CHD patients and to establish its predictive value for important cardiovascular end points while simultaneously controlling for potential confounders, including emerging biomarkers reflecting important aspects in CHD patients, such as inflammation [C-reactive protein (CRP), lipoprotein-associated phospholipase A2 (Lp-PLA2), and secretory phospholipase A2 (sPLA2)], hemodynamic stress (NT-proBNP), and renal impairment (cystatin C).
Materials and Methods
STUDY POPULATION
All patients with CHD [International Classification of Diseases, 9th Revision (ICD-9), codes 410–414] who were 30–70 years of age and who chose to participate in an in-hospital rehabilitation program between January 1999 and May 2000 in 2 cooperating hospitals (Schwabenland-Klinik, Isny, and Klinik im Südpark, Bad Nauheim, Germany) were enrolled in the study (initial response, 58%). In Germany, all patients who experience an ACS event or undergo elective CABG are offered a comprehensive in-hospital rehabilitation program after their discharge from the acute-care hospital. The aim of this 3-week program is to reduce cardiovascular risk factors, improve the health-related quality of life, and preserve the ability to work (the last only if a participant was working at the time of disease onset; otherwise, the goal was to prevent nursing care). This in-hospital rehabilitation program usually starts approximately 3 weeks after the acute event or CABG. In the current study, we included only patients who were admitted within 3 months of the acute event or CABG.
All participants gave written informed consent. The study was approved by the ethics boards of the Universities of Ulm and Heidelberg and by the physicians' chamber of the states of Baden-Wuerttemberg and Hessen (Germany).
DATA COLLECTION
At the beginning of the in-hospital rehabilitation program, all participants filled out a standardized questionnaire regarding sociodemographic information and medical history. In addition, information was taken from the patients' hospital charts. For all patients, active follow-up was conducted 1, 3, 4.5, 6, and 8 years after discharge from the rehabilitation center. Information was obtained from the patients via a mailed standardized questionnaire. Information regarding secondary cardiovascular disease (CVD) events and treatment since discharge from the in-hospital rehabilitation clinic was obtained from the primary care physicians, also by means of a standardized questionnaire. If a participant had died during follow-up, the death certificate was obtained from the local public health department, and the main cause of death was coded according to ICD-9 codes 390–459 and ICD-10 codes I0-I99 and R57.0. Secondary CVD events were defined as CVD as the main cause of death (as stated in the death certificate), as nonfatal MI, or as nonfatal stroke. All nonfatal secondary events were reported by the primary care physicians.
LABORATORY METHODS
At baseline, blood was drawn at the time of discharge from the rehabilitation center (mean, 43 days after the acute event; first quartile, 36 days; third quartile, 51 days) with the patient in a fasting state and under standardized conditions, and blood samples were stored at −80 °C until analysis. All hs-cTnT measurements were made with a high-sensitivity assay (Roche Diagnostics) on an Elecsys 2010 platform (Roche). A value of 14 ng/L has recently been reported to represent the 99th percentile in a healthy reference population (6). The interassay CVs were 3.6% and 2.9% at concentrations of 42 and 2.82 ng/L, respectively. CRP concentrations were measured by immunonephelometry on a Dade Behring Nephelometer II (N Latex CRP mono assay; Dade Behring). Cystatin C was measured on the same device (18). NT-proBNP was measured by electrochemiluminescence on an Elecsys 170 instrument (Roche Diagnostics) (19). Lp-PLA2 and sPLA2 mass and activity were measured as previously reported (20, 21).
Interassay CVs were 4.1% for CRP, 3.8% for cystatin C, and between 3% (for a target value of 4970 pg/mL) and 7% (for a target value of 178 pg/mL) for NT-proBNP. All markers were measured in a blinded fashion. Creatinine, blood lipids, and the leukocyte count were measured by routine methods in both participating hospitals.
STATISTICAL METHODS
The study population was described with respect to various sociodemographic and medical characteristics. The associations of sociodemographic characteristics, various cardiovascular risk factors, and medications with hs-cTnT concentration were quantified by means of the nonparametric Kruskal–Wallis test. Partial Spearman correlation coefficients, adjusted for age and sex, were calculated for hs-cTnT concentrations and blood lipids, CRP, creatinine clearance, cystatin C, NT-proBNP, Lp-PLA2, and sPLA2.
The relationships of hs-cTnT concentration with CVD events during follow-up were assessed by the Kaplan–Meier and life-table method and quantified by means of the log-rank test. We then used modeling with Cox proportional hazards regression analysis to assess the independent association of hs-cTnT concentration with the risk of secondary CVD events [hazard ratios (HRs) and their 95% CIs]. A basic model was adjusted for age (in years) and sex. In a second step, potential confounding variables were added to the model with the main factor (hs-cTnT) and the variables of age and sex. The following potential confounders were considered in multivariable analyses: body mass index (in kilograms per square meter), smoking status (never, current, ex-smoker), duration of school education (<10 years, ≥10 years), hospital site (Isny, Bad Nauheim), family status (married, other), history of MI (yes, no), history of hypertension (yes, no), history of diabetes mellitus (yes, no), severity of CHD (number of affected epicardial coronary vessels at baseline), initial management of CHD (conservative, percutaneous coronary intervention, CABG, intake of β-blockers) (yes, no), intake of angiotensin-converting enzyme inhibitors (yes, no), intake of diuretics (yes, no), intake of lipid-lowering drugs (yes, no), HDL cholesterol (in milligrams per deciliter), and LDL cholesterol (in milligrams per deciliter). To avoid overadjustment, we added only variables to the model that predicted a secondary event at an α level of 0.1 or that changed the parameter estimates for the main variables (hs-cTnT) by >10%. We performed a final modeling step to adjust for CRP, cystatin C, and NT-proBNP.
Measures of model fit, discrimination, reclassification, and calibration were assessed with Cox proportional hazards regression. Goodness of model fit was assessed statistically with likelihood ratio, Akaike information criterion, and Bayesian information criterion measures. The net reclassification improvement by adding hs-cTnT was calculated according to the risk strata of 5%, 10%, and >20% of the predicted probability for a cardiovascular event. Furthermore, we calculated the integrated discrimination improvement, which estimates the extended model's improvement in the difference in predicted probabilities between cases and noncases (22, 23). To further demonstrate the prognostic value of hs-cTnT, we used a modified GRACE score (13), which replaced the information about ST-segment depression with information of ischemic signs obtained during an exercise stress test.
All statistical procedures were carried out with the SAS statistical software package (release 8.2; SAS Institute).
Results
The main sociodemographic and laboratory characteristics of the 1050 patients with clinically manifest CHD are presented in Table 1. The mean age of these CHD patients (85.1% men) was 58.9 years. The patients' main laboratory characteristics are also summarized in this table. The median hs-cTnT concentration was 10.9 ng/L [interquartile range (IQR), 5.1–18.9 ng/L], 84% of all patients had hs-cTnT concentrations >3 ng/L (the lower detection limit), and 37.1% (n = 389) had concentrations greater than the reference population 99th percentile cutoff of 14 ng/L. The mean LDL cholesterol concentration was 100.5 mg/dL (2.60 mmol/L), with most patients being on a statin drug.
The distribution of hs-cTnT concentrations with respect to various factors (Table 2) indicates that hs-cTnT values differed between the sexes, were age dependent, and were higher in diabetic patients (defined by history), in hypertensive patients, in patients with multivessel CHD (>1 affected epicardial artery), in patients with an impaired left ventricular function, and in patients after CABG. In addition, hs-cTnT concentrations were higher in patients with sinus tachycardia, in those with atrial dysrhythmias, and in those with anterior wall infarction (Table 2).
Partial Spearman coefficients of the rank correlation between the hs-cTnT concentration and various other biomarkers given in Table 3 indicate that hs-cTnT concentration was positively correlated with leukocyte count, CRP, interleukin-6, adiponectin, cystatin C, NT-proBNP, Lp-PLA2 mass and activity, and sPLA2 mass and activity (all P values were statistically significant). The strongest correlations, however, were seen with NT-proBNP (ρ = 0.61) and cystatin C (ρ = 0.32).
During a median follow-up of 8.1 years (IQR, 6.1–8.2 years), 150 patients (14.3%) experienced a secondary CVD event (cardiovascular death, 39%; nonfatal MI, 33%; nonfatal stroke, 27%). The baseline hs-cTnT concentration was higher in participants with an event than in those who were event free [median hs-cTnT concentrations of 15.7 ng/L (IQR, 9.4–25.3 ng/L) and 10.0 ng/L (IQR, 4.7–17.5 ng/L), respectively; P < 0.001].
Kaplan–Meier curves of the incidence of patients with secondary CVD events per 1000 patient-years according to hs-cTnT quartile at baseline are shown in Fig. 1. Of note is that of the patients in the top hs-cTnT quartile, 37.5 experienced an event per 1000 patient-years, compared with only 11.4 per 1000 patient-years in the bottom quartile (P = 0.0001).
Kaplan–Meier curves for hs-cTnT quartiles are indicated as follows: 1 (—M), 2 (— –), 3 (- - - -), 4 (– – –). Incidence is per 1000 patient-years (*).
The association of hs-cTnT with the composite end point after adjustment for various confounders is given in Table 4. In the multivariate model that included (in addition to age, sex, and hospital site) smoking status, history of diabetes, initial management of CHD, LDL-C, HDL-C, and treatment with lipid-lowering drugs (>90% of which were statins), hs-cTnT concentration in the top quartile was associated with an HR of 2.83 (95% CI, 1.68–4.79), compared with the bottom quartile. After further adjustment for other emerging and established biomarkers of subsequent CVD risk, such as CRP, cystatin C, and NT-proBNP, the HR was slightly attenuated (2.27; 95% CI, 1.31–3.95). Analyzing hs-cTnT as a continuous variable produced only minor changes in the risk estimates across all models, and the results remained statistically significant in the final model.
Finally, various measures of model accuracy with and without hs-cTnT demonstrated (Table 5) that, in all models, the addition of hs-cTnT significantly improved model fit. A small but nonsignificant increase was seen in the area under the ROC curve from 0.67 (95% CI, 0.627–0.714) to 0.69 (95% CI, 0.647–0.733). The net reclassification improvement was 17.2% (P = 0.029) when hs-cTnT was added to the model. The integrated discrimination improvement was 0.009 (P < 0.0001). On the basis of a model of a modified GRACE score, adding hs-cTnT to the model significantly improved the prognostic value [net reclassification improvement, 19.2% (P = 0.0009); integrated discrimination improvement, 0.015 (P = 0.0002)].
Discussion
We measured troponin with a high-sensitivity assay in a large well-characterized cohort of patients with stable CHD. Our data clearly demonstrate that slight increases in hs-cTnT concentration in this population are associated with a substantial increase in the risk for hard cardiovascular end points during long-term follow-up (median, 8.1 years). Even after extensive adjustments for a variety of potential confounders, including such emerging biomarkers as cystatin C and NT-proBNP, the association was only marginally attenuated. The risk was still increased by greater than 2-fold when the extreme quartiles were compared.
The median hs-cTnT concentration was 10.9 ng/L, which is below the 99th percentile cutoff of 14 ng/L for a healthy population. More than 80% of the patients had a detectable concentration, and 37% had concentrations >14 ng/L. Omland et al. (16) studied 3679 patients with stable CHD and a preserved left ventricular function and found similar distributions. These authors also assessed determinants for increased hs-cTnT concentrations in participants of their study. They reported strong associations of hs-cTnT with age, body mass index, history of percutaneous coronary intervention, CABG, stroke, left ventricular function, diabetes, hypertension (blood pressure), smoking, impaired renal function, and several drug regimens (including aspirin, lipid-lowering drugs, diuretics, and digitalis). In addition, they found strong correlations of hs-cTnT with CRP and with NT-proBNP. We confirmed most of these associations in our study, except for the association with smoking, and we have extended the observations of Omland et al. by identifying associations with several additional variables that they had not reported. Slightly increased hs-cTnT concentrations in our study were also associated with the severity and extent of coronary artery disease, the localization of the index MI in the anterior wall, and various rhythm disturbances, such as sinus tachycardia and atrial flutter or fibrillation. The mechanisms for the release of cTnT in this population might include transient ischemia, hemodynamic stress (as occur during large MIs), hypertension, both systolic and diastolic blood pressures, diabetes, impaired renal function, and an ongoing increased inflammatory response after the acute event (24). Some of the aforementioned mechanisms are controversial, however (25). A rapid ventricular rhythm also may lead to myocyte necrosis (26).
Even very low hs-cTnT concentrations strongly predicted the occurrence of fatal and recurrent nonfatal CVD events in this cohort of patients with clinically manifest stable CHD following an acute event. Notably, this increase was not related to the elapsed time since the acute event and was independent of the type of initial management. The increase was most pronounced in the top quartile of the hs-cTnT distribution (>18.6 ng/L), which had 37.5 events per 1000 patient-years, compared with 11.4–21.3 events per 1000 patient-years in quartiles 1–3. The HR comparing the extreme quartiles as assessed by a Cox proportional hazards model was 2.27 (95% CI, 1.31–3.95) after multiple adjustments. A similar incidence was found in the PEACE trial for cardiovascular death, but no association was found with nonfatal MI (16). In our study, fatal cardiovascular events were relatively rare; additional subgroup analyses unfortunately were not feasible because of the relatively small number of events.
Measures of model accuracy indicated that the addition of hs-cTnT to a basic model improved model fit (increase in the likelihood ratio, decreases in the Akaike and Bayesian information criteria; Table 5) and nonsignificantly improved the area under the curve in ROC curve analyses (from 0.67 to 0.69). Reclassification indices also improved slightly, and model calibration was good; however, although these changes in model fit, discrimination, and reclassification were in the right direction, the improvements were modest. Similar improvements were found by Omland et al. (16).
Our study has several limitations that need to be addressed. Although we had a large sample of patients with CHD (>50% with a history of MI), fatal CVD events were limited in this study population. An explanation for this fact is that mortality due to MI is highest during the prehospital and early in-hospital phase. Because the acute events that led to a diagnosis of CHD or MI had occurred at least 3 weeks before patients were included in this study, we must assume that the patient population selected had a better prognosis than a patient population in the early phase of newly diagnosed CHD. We chose to include only “hard” secondary CVD events (such as MI, stroke, and fatal cardiovascular events) as outcomes and did not consider “softer” end points, such as a need for revascularization or nonfatal hospitalizations for heart failure, because the latter category may also be determined by healthcare use and other factors related to characteristics of specific providers. Furthermore, not all patients were willing or able to participate in an in-hospital rehabilitation program. That may be an additional explanation for the underrepresentation of severely ill patients in our study sample; however, it does not explain the positive association between hs-cTnT concentration and CVD events and suggests that the true prognostic value of hs-cTnT is even stronger than our study has shown.
In conclusion, the slightly increased but very low hs-cTnT concentrations in patients with stable CHD strongly predicted recurrent cardiovascular events during long-term follow-up, even after extensive adjustments for multiple covariates. The degree of model improvement after adding hs-cTnT to a basic clinical model was modest, however. Thus, the clinical utility of routine hs-cTnT measurements in stable CHD patients is still questionable. Before widespread measurement of troponins with high-sensitivity assays can be recommended for patients with stable CHD, larger studies with contemporary, unselected patients are needed to clearly demonstrate a clinically relevant improvement in risk prediction after the addition of hs-cTnT to a model with standard clinical and cardiometabolic risk factors.
Footnotes
-
↵6 Nonstandard abbreviations:
- ACS,
- acute coronary syndrome;
- MI,
- myocardial infarction;
- hs-cTnT,
- cardiac troponin T as measured with a high-sensitivity assay;
- CHD,
- coronary heart disease;
- TIMI,
- Thrombolysis in Myocardial Infarction (score);
- GRACE,
- Global Registry of Acute Coronary Events (score);
- CABG,
- coronary artery bypass grafting;
- PEACE,
- Prevention of Events with Angiotensin-Converting Enzyme Inhibition (trial);
- HOPE,
- Heart Outcomes Prevention Evaluation (study);
- NT-proBNP,
- N-terminal pro–B-type natriuretic peptide;
- CRP,
- C-reactive protein;
- Lp-PLA2,
- lipoprotein-associated phospholipase A2;
- sPLA2,
- secretory phospholipase A2;
- ICD-9,
- International Classification of Diseases, 9th Revision;
- CVD,
- cardiovascular disease;
- HR,
- hazard ratio;
- IQR,
- interquartile range.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: No authors declared any potential conflicts of interest.
Role of Sponsor: No sponsor was declared.
- Received for publication January 30, 2012.
- Accepted for publication May 14, 2012.
- © 2012 The American Association for Clinical Chemistry
References
