Abstract
BACKGROUND: High-sensitivity cardiac troponin (hs-cTn) T and I assays are established as crucial tools for the diagnosis of acute myocardial infarction (AMI), as they have been found superior to old troponin assays. However, eventual differences between the assays in prediction of significant coronary lesions and long-term prognosis in patients with acute coronary syndrome (ACS) have not been fully unraveled.
METHODS: Serum concentrations of hs-cTnT (Roche), hs-cTnI (Abbott), and amino-terminal pro-B-type natriuretic peptide (NT-proBNP; Roche) in 390 non-ST-elevation (NSTE) ACS patients were evaluated in relation to significant coronary lesions on coronary angiography (defined as a stenosis >50% of the luminal diameter, with need for revascularization) and prognostic accuracy for cardiovascular mortality, all-cause mortality, as well as the composite end point of cardiovascular mortality and hospitalizations for AMI or heart failure.
RESULTS: The mean (SD) follow-up was 2921 (168) days. Absolute hs-cTnI concentrations were significantly higher than the hs-cTnT concentrations. The relationship between analyzed biomarkers and significant coronary lesions on coronary angiography, as quantified by the area under the ROC curve (AUC), revealed no difference between hs-cTnT [AUC, 0.81; 95% CI, 0.77–0.86] and hs-cTnI (AUC, 0.81; 95% CI, 0.76–0.86; P = NS). NT-proBNP was superior to both hs-cTn assays regarding prognostic accuracy for both cardiovascular and all-cause mortality and for the composite end point during follow-up, also in multivariate analyses.
CONCLUSIONS: The hs-cTnT and hs-cTnI assays displayed a similar ability to predict significant coronary lesions in NSTE-ACS patients. NT-proBNP was superior to both hs-cTn assays as a marker of long-term prognosis in this patient group.
The definition of acute myocardial infarction (AMI)6 continues to evolve as more sensitive biomarkers are developed (1). Measurement of cardiac troponins (cTns) has become important in the diagnosis of AMI (2), and rapid identification of AMI is essential for the initiation of effective treatment and management (1, 3).
High-sensitivity cardiac troponin (hs-cTn) assays have recently been shown to be superior to conventional cTn assays for the early diagnosis of AMI (4, 5). Current guidelines recommend the use of a cTn cutoff-concentration at the 99th percentile of healthy controls, together with a rise and/or fall pattern (1, 2, 6). The measurement of hs-cTn should be made at admission and repeated 3–6 h later (2, 7, 8). Owing to increased but stable concentrations of cTn in some chronic diseases, dynamic changes become a critical part of the AMI diagnosis (9, 10). However, it is essential to separate dynamic changes resulting from AMI from dynamic changes associated with analytical imprecision or normal chronobiological variation (1, 11).
Only 2 hs-cTn assays are currently available for clinical use in Europe [hs-cardiac troponin T (cTnT) (Roche Diagnostics) and hs-cardiac troponin I (cTnI) (Abbott Laboratories)]. Studies have unveiled that both assays have high accuracy in early diagnosis of AMI, as well as a high negative predictive value for acute coronary syndrome (ACS) (4, 12, 13, 14). Since both assays are in clinical use, direct comparison of the two is of considerable clinical importance, as differences might be relevant for clinical utility (15). Accordingly, the present study aimed to directly compare the performance of hs-cTnT and hs-cTnI as biomarkers of clinically significant coronary lesions and long-term prognosis in patients with non-ST elevation ACS (NSTE-ACS). Further, we wanted to elucidate if differences in performance between the 2 assays most likely relate to biological differences between the troponins or analytical differences between the assays. Amino-terminal pro–B-type natriuretic peptide (NT-proBNP) was included for comparison as a prognostic marker, and contemporary sensitive cTn assays were included for assessment of covariation between different cTn assays.
Materials and Methods
STUDY DESIGN AND POPULATION
This clinical study was conducted between 2007 and 2008, in a single coronary care center, including 458 consecutive patients admitted for coronary angiography due to NSTE-ACS, and the methods have been described previously in detail (16). The classification of patients as non-ST elevation myocardial infarction (NSTEMI) or unstable angina pectoris (UAP) was based on clinical judgment, electrocardiography (ECG) recordings, and rise and/or fall of cTn concentrations as measured by contemporary sensitive assays (both cTnT and cTnI) used in 2007–2008 in the hospitals in the catchment area of the coronary care center.
The present study includes 390 of the original population, with remaining serum samples available for further analysis. History of coronary artery disease (CAD) was predefined as previous AMI and/or coronary revascularization, either by percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG). ECG recordings were analyzed by one investigator (JG) with no knowledge of the hs-cTnT nor hs-cTnI concentration at the time of the ECG assessment. The Regional Ethics Committee approved the study, and all patients provided written, informed consent.
BIOCHEMICAL ANALYSIS
Peripheral venous blood was collected during inclusion, which occurred within 2 h before coronary angiography. After centrifugation, serum aliquots were frozen and stored at −70 °C until analysis. All biomarkers presented in the study are from samples drawn at the same time point. The Roche cTnT (fourth generation), hs-cTnT, and NT-proBNP assays were performed on a Modular E170 platform using the Elecsys reagents. Whereas the fourth-generation cTnT assay had a limit of detection (LoD) of 10 ng/l and a 10% CV at 35 ng/L, the hs-cTnT assay had a limit of blank of 3 ng/L, a LoD of 5 ng/L, a 10% CV at 13 ng/L, and a 99th percentile in healthy individuals reported to be 14 ng/L (17). NT-proBNP had a LoD of 5 ng/L, and a 97.5th-percentile cutoff of 263 ng/L for those ages 55–64. The interassay CV was 3.1% at a concentration of 46 ng/L and 2.7% at a concentration of 125 ng/L. The Abbott hs-cTnI assay (ARCHITECT STAT high-sensitive Troponin I) had a LoD of 1.9 ng/L, a 99th percentile in healthy individuals of 26 ng/L, and 10% CV at 4.7 ng/L (15, 18). The Siemens cTnI Ultra assay had a LoD of 6 ng/L and a 99th percentile of 40 ng/L, and samples were analyzed as previously reported (16). To exclude the possibility that different timing from sampling to analysis could have affected the results (hs-cTnT analyzed in 2010 and hs-cTnI analyzed in 2014), 40 samples from all quartiles were reanalyzed for hs-cTnT in May 2016. These results displayed no deterioration of sample quality after longtime storage (see Fig. 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol63/issue2.). The investigational assays were all commercially available and supplied by the respective manufacturers, which had no role in the design of the study, the analysis of the data, or the preparation of the manuscript. Renal function was evaluated by the estimated glomerular filtration rate (eGFR), as recommended by the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) (19). Renal dysfunction was defined as eGFR <60 mL · min−1 · (1.73 m2)−1.
CORONARY ANGIOGRAPHY
Coronary angiography was performed by a standard (Judkins) technique, using digital image acquisition and storage. Revascularization was not part of the study protocol but was performed in accordance with current guidelines.
STUDY OUTCOMES
The primary end point was significant coronary lesions, defined as a coronary artery stenosis >50% of the luminal diameter with a need for revascularization, either by PCI or CABG, or significant lesions not suitable for PCI or CABG. The secondary end points evaluating long-term prognosis were (a) cardiovascular mortality, defined as time to death from an acute cardiovascular event or sudden death without other obvious cause; (b) all-cause mortality, defined as time to death irrespective of cause; and (c) composite cardiovascular end point, defined as the composite of cardiovascular mortality and hospitalization for recurrent AMI or heart failure (HF). Events were registered in all patients.
STATISTICAL ANALYSIS
Continuous variables with a normal distribution are described with means and standard deviation (SD), and variables with a skewed distribution with median and quartiles. Categorical and discrete variables are presented as counts and percentages. Paired differences in continuous variables were tested using the Wilcoxon test, and association between pairs of categorical variables was tested using χ2 tests with Yates correction. The unadjusted accuracy of the respective biomarkers in prediction of significant coronary lesions and long-term prognosis was determined by area under the ROC curves (AUCs). Survival times were defined from the date of diagnosis to death, as previously defined, and presented through Kaplan–Meier curves. Continuous, log-transformed values of hs-cTn and NT-proBNP were included in a Cox regression analysis to evaluate their associations with the respective end points. We adjusted for age and history of CAD (step 1) and for log-transformed plasma concentrations of C-reactive protein (CRP) and eGFR (step 2), as these are factors associated with increased cTn-concentrations (9). Hazard ratios (HRs) and 95% CIs were calculated. The relationships between the different cTn assays were examined on log-transformed values by Spearman rank correlation and a Deming regression method for first- and second-order polynomials. The sum of squared residuals from the actual data point orthogonal to the regression curve was minimized by a bootstrap method. These regression residuals also constitute the basis for calculation of intermethod dispersion (IMD) = 10δ − 1, where δ is the SD of the Deming residuals. Hence, IMD expresses how much the ratio between the 2 methods (10δ) deviates from 1 and, accordingly, denotes the dispersion between the log-transformed cTn values. All statistical tests were 2-sided, and a significance level of 0.05 was used. We consider our analysis as exploratory and therefore did not adjust the significance level for multiple testing. Comparison between AUCs was performed as previously described, using MedCalc software version 14.12.0 and the DeLong test (20). All other statistical analyses were performed using SPSS version 22.0 (SPSS Inc.).
The authors are solely responsible for the design and conduct of this study, all study analysis, and drafting and editing of the manuscript.
Results
PATIENT CHARACTERISTICS
Baseline characteristics for the total patient cohort are shown in Table 1. The median period from onset of symptoms to coronary angiography was 2 days. There were 41 (10.5%) patients with renal dysfunction, defined by eGFR <60 mL · min−1 · (1.73 m2)−1, included in the study. Positive angiographic findings were detected in 282 patients (72%), with one or more significant coronary lesions. In this group, 198 patients (70%) underwent PCI and 62 patients (22%) underwent CABG. The remaining 22 patients (8%) had significant coronary lesions not suitable for revascularization. In the patient group with UAP (n = 118) and the patient group with NSTEMI (n = 272), positive angiographic findings were detected in 45% and 86% of the patients, respectively.
Baseline characteristics.
COMPARISON OF hs-cTnT AND hs-cTnI CONCENTRATIONS
Neither hs-cTnT nor hs-cTnI concentrations were significantly influenced by the history of CAD (Fig. 1A). Both hs-cTnT and hs-cTnI concentrations were higher among patients with NSTEMI vs patients with UAP (Fig. 1B), and among nonsurvivors vs survivors (Fig. 1C and online Supplemental Fig. 2A) during follow-up. The respective differences in hs-cTnT and hs-cTnI concentrations between survivors and nonsurvivors were also present after exclusion of patients with renal dysfunction (Fig. 1D). In general, hs-cTnI concentrations were numerically higher than hs-cTnT values. The correlation coefficients between hs-cTnT and hs-cTnI were 0.965 (P < 0.001) for the whole patient group, and 0.699 (P < 0.001) and 0.947 (P < 0.001) in the UAP and NSTEMI groups, respectively. The correlation coefficient for the UAP group remained significantly lower than for the NSTEMI group after exclusion of patients with renal dysfunction (0.690 vs 0.952, P < 0.001). There was no significant difference in variables that predicted increasing concentrations of hs-cTnT than of hs-cTnI (data not shown).
COMPARISON OF THE ABILITY TO PREDICT SIGNIFICANT CORONARY LESIONS
The unadjusted test accuracy of one single measurement of hs-cTnT or hs-cTnI for identification of significant coronary lesions in the total patient cohort was similar between the 2 assays (Fig. 2A, Table 2). These results were comparable for the subgroup of patients without renal dysfunction (see online Supplemental Table 1 and online Supplemental Fig. 3A). The sensitivities, specificities, negative predictive values, and positive predictive values at LoD and the 99th percentiles of the assays are presented in online Supplemental Table 2.
Kaplan–Meier curves demonstrating the cumulative incidence of cardiovascular mortality in the total patient cohort according to quartiles of hs-cTnT (C) and hs-cTnI concentrations (D), as well as cardiovascular mortality in the subgroup of patients without renal dysfunction according to quartiles of hs-cTnT (E) and hs-cTnI concentrations (F).
Prediction of significant coronary lesions and long-term prognosis.
COMPARISON OF PROGNOSTIC PERFORMANCE
The mean (SD) follow-up period was 2921 (168) days. The unadjusted accuracy of a single measurement of hs-cTnT, hs-cTnI, or NT-proBNP in prediction of cardiovascular mortality (n = 34), all-cause mortality (n = 51), and the composite end point (n = 92) was tested by respective AUCs. NT-proBNP was superior to both hs-cTn assays in prediction of all secondary end points (Table 2, Fig. 2B, and online Supplemental Fig. 2B). When comparing the 2 hs-cTn assays, a minor but statistically significant difference in favor of hs-cTnT was detected (Table 2). These differences were essentially unaltered when separate analyses on the subgroup of patients without renal dysfunction were performed (see online Supplemental Table 1 and online Supplemental Fig. 3B).
Log-transformed serum concentrations of both hs-cTn assays and NT-proBNP were separately included in a univariate Cox proportional hazards regression model (Table 3). Whereas significant associations were found between all 3 assays and cardiovascular, as well as all-cause mortality during follow-up, only NT-proBNP was significantly associated with the composite cardiovascular end point. In contrast to the hs-cTnT and hs-cTnI assays, NT-proBNP remained a significant predictor for both mortality and the composite cardiovascular end point after adjustment for age, history of CAD, log-transformed plasma concentrations of CRP (missing data in n = 69 patients), and eGFR (Table 3, regression step 2). Multivariate analyses for the subgroup of patients without renal dysfunction are presented in online Supplemental Table 3. Kaplan–Meier curves depicting the cumulative incidence of cardiovascular mortality in all patients according to quartiles of the respective biomarker concentrations are shown in Fig. 2, C and D, whereas the corresponding analyses for all-cause mortality are displayed in online Supplemental Fig. 2, C and D. Both hs-cTnT and hs-cTnI concentrations in the fourth quartile seem to identify a subgroup of patients with a particularly poor prognosis. These findings remained comparable within the subgroup of patients with eGFR ≥60 mL · min−1 · (1.73 m2)−1, and therefore could not be explained by the contribution of renal dysfunction alone (Fig. 2, E and F).
Multivariate analysis; effects of hs-cTnT, hs-cTnI, and NT-proBNP concentrations on long-term prognosis.
VARIATION BETWEEN DIFFERENT TROPONIN ASSAYS
Fig. 3 shows scatterplots for comparisons between log-transformed cTn concentrations, measured with both new and old assays. Surprisingly, the observed relationships between the different cTnT and cTnI assays were nonlinear. To quantify the dispersion between the different cTn assays, IMD was calculated as detailed previously in the Materials and Methods section. The estimated IMD was 9.4% between the cTnT methods, whereas it was 16.1% between the cTnI methods. However, when comparing hs-cTnT with the hs-cTnI and Siemens cTnI Ultra methods, the IMD increased markedly and varied between 45.7% and 48.6% (Fig. 3).
IMDs, expressed in relative units, are displayed in the corresponding retransformed figures.
Discussion
This study, which directly compared the 2 commercially available hs-cTn assays (Roche hs-cTnT and Abbott hs-cTnI) in NSTE-ACS patients, revealed novel and complementary findings on the following topics: (a) prediction of significant coronary lesions, (b) long-term prognosis during a mean follow-up of more than 8 years, and (c) thorough analyses of the relationship and dispersion between different cTn methods.
When comparing one single measurement of hs-cTnT and hs-cTnI for the identification of significant coronary lesions, we found similar accuracy between the 2 assays. The long-term prognostic performance of NT-proBNP was superior to both hs-cTnT and hs-cTnI, also within the subgroup of patients without renal dysfunction. Although hardly relevant clinically, a small but statistically significant difference in prognostic accuracy between the 2 hs-cTn assays, in favor of hs-cTnT, was detected. When assessing the relationships between log-transformed values from the different cTn-methods, nonlinear relationships were seen. We found only a minor difference in dispersion between the hs-cTnT and the contemporary cTnT assays and between the hs-cTnI and the contemporary cTnI assays, respectively. Conversely, a striking finding was the large dispersion between the hs-cTnT and the cTnI assays.
Since both hs-TnT and hs-TnI assays are in clinical use as markers of myocardial injury, NSTE-ACS patients may be referred from emergency hospitals to invasive coronary care centers that use different hs-cTn methodology. Hence, studies comparing decision limits and prognostic performance between different hs-cTn assays are of clinical importance. Given the fact that these are 2 different proteins and several different analytical methods exist, both biological and analytical properties are possible factors that may explain the observed differences. A recent report directly compared the hs-cTnT and hs-cTnI assays for the early diagnosis of AMI among unselected patients admitted with acute chest pain (15). Although subtle, interesting differences were found. The diagnostic performance of hs-cTnI at admission was superior to that of hs-cTnT among the subgroup of early presenters. However, there was no difference between the assays within the total cohort. Analogous to our results, hs-cTnT was found to be superior to hs-cTnI in prediction of long-term prognosis.
As the sensitivity of the cTn assays has improved, new information has emerged indicating that troponins are also important prognostic biomarkers in cardiovascular disorders other than AMI (21). The strong prognostic value of increased troponin concentrations as measured with high-sensitivity assays has been demonstrated in chronic nonischemic cardiac conditions such as atrial fibrillation, HF, and aortic stenosis (22–25). Even within a low-level range, cTns measured with both hs-cTnT and hs-cTnI assays were important predictors for cardiovascular outcomes in patients with stable angina pectoris (26, 27). In our opinion, these studies demonstrate that increased cTn concentrations are of considerable pathophysiological importance beyond what is related to ischemia directly. In this respect, changes in hs-cTnT and hs-cTnI concentrations in patients referred for myocardial perfusion imaging owing to suspected stable CAD were more highly correlated with variables associated with structural myocardial alterations than with reversible ischemia (28). A short-term biological variation in cTn concentrations has also been shown for healthy individuals and in patients with stable CAD (29, 30). In the present study, the correlation coefficient between hs-cTnT and hs-cTnI concentrations was substantially lower within the UAP subgroup than among NSTEMI patients, irrespective of exclusion of those with renal dysfunction. Accordingly, the clinical relevance of differences between hs-cTnT and hs-cTnI assays is likely to be most prominent in patients with cTn-concentrations in the lower range, which is understandable because substantial myocardial necrosis in AMI patients will override more subtle biological or analytical differences. A poor correlation between cTnT and cTnI concentrations in the low-level range is also demonstrated in patients with nonischemic conditions (27, 31, 32, 33).
Although the pathophysiological mechanisms of cTn release from cardiac myocytes in the lower range are not completely understood, differences between cTnT and cTnI concentrations may involve differences in protein weight, different stability in the circulation, and different ratios of free cytoplasmic cTnT and cTnI of the damaged myocytes (32, 34). Interestingly, hs-cTnT concentrations have been shown to be more influenced by renal dysfunction than hs-cTnI concentrations (28). In our study, the drop in predictive value for the hs-cTn assays in step 2 of the multivariate analyses indicates an association with both renal dysfunction and inflammation. These may all be important factors responsible for the differences observed in the present study and other studies (13–15). The differences in IMD between the 2 cTnT and the 2 cTnI assays, respectively, could be explained by calibration/standardization of the assays. While the cTnT assays are available only from Roche and therefore use the same monoclonal antibodies, there are several cTnI manufacturers using different detection technology (32). However, the observation that the dispersion is dramatically larger between cTnT and cTnI assays than between the 2 cTnI assays is unlikely to be explained by analytical differences alone. This suggests that there are biological differences between the troponins that also influence measurements, which in turn may affect prognostic value. Thus, differences between cTnT and cTnI could be of clinical relevance for prognostic utility, but these prognostic differences do not seem to influence the ability of these assays to identify significant coronary artery lesions in NSTE-ACS patients.
LIMITATIONS
As our results in this exploratory study are based on one single measurement of cTn after admittance to a tertiary coronary center, we cannot exclude the possibility that earlier or serial sampling would have affected the results. Our data do not take into consideration potential differences in performance among early presenters. Although the time intervals from symptoms to coronary angiography were registered, the precise time point for sampling was not further specified than within 2 h before coronary angiography. We cannot exclude a selection bias in our study, with a relatively higher number of patients with significant CAD, because the patients were referred to coronary angiography during the inclusion period based on troponin concentrations measured by contemporary sensitive cTn assays.
Conclusions
The hs-cTnT and hs-cTnI assays displayed a similar ability to predict significant coronary lesions among NSTE-ACS patients. NT-proBNP is superior to both hs-cTn assays as a prognostic marker in this patient group. Our data suggest that the observed differences between hs-cTnT and hs-cTnI assays have biological rather than analytical explanations. However, further studies are needed to further explore the relevance of differences between the 2 high-sensitivity troponin assays in cardiovascular disease.
Acknowledgments
The authors highly appreciate the technical assistance of Marit Holmefjord Pedersen, Clinical Trail Unit, Division of Medicine, Akershus University Hospital.
Footnotes
↵6 Nonstandard abbreviations:
- AMI,
- acute myocardial infarction;
- cTns,
- cardiac troponins;
- ACS,
- acute coronary syndrome;
- NSTE,
- non-ST elevation;
- NT-proBNP,
- amino-terminal pro–B-type natriuretic peptide;
- NSTEMI,
- non-ST elevation myocardial infarction;
- UAP,
- unstable angina pectoris;
- ECG,
- electrocardiography;
- CAD,
- coronary artery disease;
- PCI,
- percutaneous coronary intervention;
- CABG,
- coronary artery bypass grafting;
- LoD,
- limit of detection;
- eGFR,
- estimated glomerular filtration rate;
- CKD-EPI,
- Chronic Kidney Disease Epidemiology Collaboration;
- HF,
- heart failure;
- AUC,
- area under the ROC curve;
- CRP,
- C-reactive protein;
- HR,
- hazard ratio;
- IMD,
- intermethod dispersion.
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: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:
Employment or Leadership: None declared.
Consultant or Advisory Role: None declared.
Stock Ownership: None declared.
Honoraria: J. Gravning, Siemens and Abbott Laboratories.
Research Funding: None declared.
Expert Testimony: None declared.
Patents: None declared.
Role of Sponsor: No sponsor was declared.
- Received for publication May 31, 2016.
- Accepted for publication August 22, 2016.
- © 2016 American Association for Clinical Chemistry
References
In this issue
