Comparison of high sensitive and conventional troponin assays in diagnosis of acute myocardial infarction Akut Tanısında Yüksek Sensitif Ve Konvansiyonel Troponin Testlerinin

Introduction : We aimed to compare the positive predictive values (PPV) and negative predictive values (NPV) of four cardiac troponin assays in the diagnosis of AMI in Turkish population. Methods : This study is an observational comparative study, which is performed between 2012 and 2013 (527 patients). Troponin levels were measured with chemiluminescence Cobas troponin T assay, immunofluorescence Triage tro­ ponin I assay, immunofluorometric Radiometer troponin I assay and immunochromatographic Toyo troponin I assay. Results: Sensitivity and specificity of immunofluoromet­ ric assay (Radiometer) are 56.82% and 71.34%; immuno­ chromatographic assay (Toyo) are 29.4% and 88.62%, immunofluorescence assay (Triage) are 47.13% and 76.12%, chemiluminescence assay (Roche) are 60.49 and 67.42%, respectively. PPV, NPV and positive likelihood ratios (LR + ) of immunofluorometric assay (Radiometer) are 45.5%, 79.7% and 1.98, immunochromatographic assay (Toyo) are 51.5%, 75.4% and 2.58, immunofluores­ cence assay (Triage) are 46.5%, 76.6% and 1.97, chemi­ luminescence assay (Roche) are 45.8%, 78.9% and 1.86, respectively. In four assays, troponin levels were statisti­ cally significant higher in AMI positive group in compari­ son to negative group (p < 0.001 for all). Conclusion: There was no statistically significant differ­ ence between these troponin methods in comparisons of PPV and NPV in the diagnosis of AMI, but low sensitivity of Triage and Toyo assays should be considered.


Introduction
The advances in the biochemical research on discovering an ideal biomarker for the diagnosis of acute myocardial infarction (AMI) have led to the discovery of cardiac tro ponins as a gold standard marker of myocardial injury [1,2]. By courtesy of advances in the immunoassay tech niques, the 99th percentile cutoff value of cardiac tro ponins required for the diagnosis of AMI decreased with the latest available ultrasensitive cardiac troponin assays capable of measuring level as low as 0.005 ng/mL [3]. Tro ponins have both diagnostic as well as prognostic signifi cance in myocardial necrosis, but the results should be interpreted in the context of clinical history, ECG findings, and cardiac imaging techniques like echocardiography to establish the correct diagnosis [4]. The monitoring of the cardiac troponins would help to differentiate acute myocardial infarction (AMI) from many other conditions such as cardiomyopathy, heart failure, renal failure, pul monary embolism, etc [5,6]. The development of high sen sitive cardiac troponin assays has led to the exclusion of the diagnosis of AMI in emergency department (ED) with a single test on admission. The undetectable troponin (< 0.003 ng/mL) with highly sensitive cardiac troponin assays at present have a high negative predictive value for AMI in patients with chest pain at ED.
In this study, we aimed to compare the positive predictive values (PPV) and negative predictive values (NPV) of four different cardiac troponin assays includ ing a new conventional troponin assay (Turklab) in the diagnosis of AMI in Turkish population and to reveal diag nostic performance of this new conventional troponin assay (Turklab).

Subjects
This observational, comparative study was conducted in Mehmet Akif Ersoy Thoracic and Cardiovascular Surgery Training and Research Hospital, Istanbul, Turkey, between February 2012 and 2013. The study patients were randomly selected from the patients who were admitted to emer gency department with chest discomfort. Patients present ing with chest pain were screened for inclusion. Exclusion criteria were chest pain due to trauma, aged < 18 years, no chest pain within 24 h of the index ED visit, chest pain lasting < 1 min, no ECG or no troponin assay performed within 24 h of index ED visit, a clear alternative diagnosis at initial medical assessment, hemodynamic instability, advanced terminal disease, inability to communicate. We collected data from patients presenting within first 6 h of symptom onset. The patients presenting outside of the first 6 h of symptom onset were also excluded. Five hundred twenty seven patients admitted to the emergency depart ment with symptoms of acute coronary syndrome were included in the study. Clinically, angiographically and echocardiographically confirmed AMI has been revealed in 163 patients. Two groups were constituted whether acute myocardial infarction diagnosis was made or not. Troponin measurements of the patients were performed via four different troponin assays within sixth hours of symptom onset. PPV and NPV of four different troponin assays in the diagnosis of AMI were compared. The study protocol was approved by the local ethics committee and written informed consent was taken from all patients. The study was conducted in accordance with the Declaration of Helsinki, Good Clinical Practice (GCP) and International Conference on Harmonization (ICH) guidelines.

Diagnosis of AMI
Symptoms and clinical characteristics considering of acute coronary syndrome such as chest pain, syncope, significant arrhythmia, significant changes of myocar dial injury on ECG such as ST depression and elevation, wall motion abnormalities on echocardiography and angiographic criteria of AMI were used to the diagnosis of AMI. The angiographic criteria of AMI was accepted as the evidence of plaque rupture or erosion in an epi cardial coronary vessel, complex lesions consisting of plaque rupture and/or thrombus formation and culprit lesions [7]. A complex lesion was defined as an acute or recent total occlusion (dye stasis at the site) or a patent vessel with a significant lesion that was usually eccentric with either overhanging edges, abrupt shoulders, ulcera tions, and/or filling defects at or distal to the lesion indi cating intracoronary thrombus [7]. A culprit lesion was considered as the only significant lesion on angiogra phy (usually > 70% diameter stenosis) and in multives sel disease, the culprit lesion was a significant lesion in a vessel that corresponded to the new electrocardio graphic changes or wall motion abnormalities [7]. The diagnosis of AMI was made by the discretion of at least two cardiologist blinded to the laboratory measurements according to above criteria.

Troponin assays and measurements
The troponin measurements were taken within sixth hours of symptom onset in accordance with current guide lines. Venous blood samples were collected by venipunc ture in a serum separator tube and K 3  Briefly, after addition of the whole blood sample, the cells are separated from the plasma via a filter in the device, and a certain amount of plasma reacts with fluorescent antibody conjugates within the reaction chamber, and the mixture flows down the device detection lane. Complexes of the analytes and fluo rescent antibody conjugates are captured on discrete zones, producing binding assays that are specific for each analyte. The concentration of each analyte, directly proportional to the fluorescence detected, is measured by the Triage meter [11]. The reportable measuring range for the Triage (cTnI) assay device is between 0.01 and 10 μg/L [10]. 4. Toyo Turklab Conventional Troponin I (cTnI) assay: EDTA whole blood sample was used to measure tro ponin I with Toyo test device (Turklab A.S, Izmir, Turkey) which is an immunochromatographic assay. There is capture reagent immobilized test area in which whole blood sample reacts with the particles coated with anticardiac troponin I antibodies. The measuring range for the Turklab (cTnI) assay device is between 0.5 and 25 μg/L.
Glucose, HDL, LDL, total cholesterol and triglycerides were measured from serum samples via Cobas systems (Roche Diagnostic Basel, Switzerland) by using commer cial kits (Roche Diagnostic Basel, Switzerland). Complete blood count (CBC) was made from EDTA whole blood samples with BC 6800 auto analyzer (Mindray Medical International Limited, Shenzhen, China). Two levels of internal quality controls were performed for all devices.

Statistical analysis
The statistical analyzes in this study were performed via the Number Cruncher Statistical System 2007 program package (NCSS statistical software, Utah, USA). For the evaluation of the data, descriptive statistical methods were used (mean, standard deviation, median, interquar tile range). For the groups not showing normal distribu tion, KruskalWallis test in the comparisons of the groups, Dunn's multiple comparison test in the subgroup com parisons, MannWhitney Utest in the comparison of the two groups were used. For the groups showing normal distribution, independent ttest in the comparison of the two groups, Spearman correlation test to determine the relationships of the variables with each other were used. In the receiver operating characteristic (ROC) analysis, the area under the ROC curve was calculated for the troponin measurements. Cut off values were selected by youden index and ztest was used for comparing AUC of two diag nostic tests in a paired design. Sensitivity, specificity, positive predictive value, likelihood ratio (LR + ) and cut points were calculated. Results were evaluated as signifi cant p < 0.05 level at 95% confidence interval.

Results
Demographic, clinical and laboratory characteristics of the patients are presented in Table 1. The presence of dia betes mellitus, hypertension, smoking and obesity did not differ between groups. The history of coronary interven tion, coronary bypass, known peripheral artery disease and cerebrovascular accident were not significant among two groups. The group without myocardial infarction was younger than the group with myocardial infarction. The group without myocardial infarction had significantly higher triglyceride levels (p = 0.003) and higher hemato crit levels (p = 0.018), but lower glucose levels (p = 0.027) and lower white blood cell counts (p = 0.0001) in com parison to the group with myocardial infarction. There was no difference between the groups in LDL, HDL, total cholesterol, creatinine levels and platelet counts (p > 0.05) (Table1). The diagnosis of myocardial infarction was confirmed by all of the four methods ( Table 2). The male patients had higher troponin levels with all of the four methods compared. In this study, 41 patients in the group AMI (-) had ST elevations, 32 of which were asso ciated with ST elevations not meeting AMI criteria such as minimal elevations, single lead or not contagious lead elevations, six of which had early repolarization, two was pericarditis and one has previously diagnosed chronic pericardial effusion.
In comparison of four troponin methods in differ ent clinical scenarios, there were significant differences among noncardiac chest pain, acute coronary syndrome (nonSTEMI) and ST segment elevation myocardial infarc tion (STEMI) groups (p = 0.0001) ( Table 3). The troponin levels of the noncardiac chest pain group were lower than other two groups (p = 0.0001) and the troponin levels of the nonSTEMI group were also lower than the STEMI group (p = 0.0001) (  (Table 7). There were no statistically significant differences between four dif ferent troponin assays in comparisons of PPV and NPV in the diagnosis of AMI but the sensitivity of Triage and Toyo Turklab methods were lower than Radiometer and Roche methods.

Discussion
According to the results of the present study, there were no statistically significant differences between four differ ent troponin methods in comparison of PPV and NPV in the diagnosis of AMI. We have found sensitivity of Triage and Toyo Turklab methods to be lower than Radiometer and Roche methods. The strength of this study comes from the comparison of the power of frequently used tro ponin methods including Toyo Turklab Conventional Tro ponin I (cTnI) assay, which is a device recently developed in Turkey that can easily measure troponin I from EDTA whole blood sample, like Radiometer and Triage methods, in the diagnosis of AMI. According to our knowledge, this is the first study comparing the power of Turklab conven tional troponin I method with other frequently used tro ponin methods in the diagnosis of AMI.
The universal definition of AMI criteria for the diagno sis of AMI is pivotally centered on elevated troponins, in an ischemic setting, with ischemic symptoms or ischemic electrocardiographic changes, and a rise and/or fall in troponin levels [4]. The cutpoint requires a troponin level greater than the 99th percentile of a healthy popula tion as measured with an assay with an imprecision CV of ≤ 10%, and a definition of rise and/or fall is required to distinguish "normal" or chronic background troponin levels from acute changes [4]. The universal definition of AMI group recommended a 20% change from the baseline value to be diagnostic of reinfarction with the current tro ponin assays [4]. The high sensitivity assays provide rapid diagnosis of AMI and it is important that there is a rising and/or falling pattern of troponin levels to distinguish nonischemic pathophysiology from AMI. The development and standardization of high sensi tive assays is also important. The firstgeneration assay for cardiac troponin T (cTnT) used bovine cTnT as the reference material and exhibited nonspecific binding to human skeletal muscle troponin, but this problem was solved by refinement of the detection antibody in the sec ondgeneration assay and the use of recombinant human   sensitivity and the specificity between the highsensitiv ity and conventional assays were statistically significant (p < 0.01), the areas under the curves were similar for both tests carried out 3-6 h after presentation [13]. Our results are similar to this metaanalysis. The areas under the ROC curves were similar for four tests. In another metaanalysis by Sethi et al. which included in 8628 patients from fifteen studies, it was found that there was no statistically signifi cant difference in the area under the curve between high sensitive troponin (95% CI: 0.920) and conventional tro ponin (95% CI: 0.929) at the 99th percentile (p = 0.62) [14].
Our results are also consistent with this metaanalysis.
Recently, Januzzi et al. has compared the perfor mance of a commercially available sensitive Tn I (sTnI) and precommercial high sensitive TnI (hsTnI) method to conventional Tn (cTn) assays with coronary computed tomographic angiography (CTA) [15]. In comparison with cTn, 29% of ACS cases previously categorized as UAP were reclassified to acute myocardial infarction with sTnI or hsTnI. In patients with acute chest discomfort, use of sTnI and hsTnI methods led to significant improvement in the early diagnostic accuracy for ACS [15]. An hsTnI below limit of detection had 100% negative predictive value for ACS or significant coronary artery stenosis in those rand omized to CTA in their study.
The high sensitivity cardiac troponin assays have improved sensitivity, but reduced specificity in compari son with the conventional troponin assays. With repeated measurements after 6 h of symptom onset, the area under the ROC curve values are similar for the four tests in our study. It has been reported that the elevation of high sen sitive troponin levels predicts higher risk of mortality in patients with suspected ACS and normal conventional tro ponin assay results, and helps in the early identification of higherrisk patients in this setting [16]. Haaf et al. also reported that hsTnT is more accurate than hsTnI in the pre diction of longterm mortality [17]. In PEACE study, it has been suggested that hsTnT concentrations were associated with cardiovascular death and heart failure but not with myocardial infarction [18], and hsTnI concentrations are cTnT for standardization in the thirdgeneration assay [12]. The fourth generation cTnT assay uses fragment anti genbinding (FAB) of two cTnT specific mouse monoclonal antibodies in a sandwich format. Nowadays, the fourth generation cTnT assay has been considered the standard assay for the diagnosis of AMI.
In a metaanalysis, AlSaleh et al. evaluated 9186 patients from 9 studies that assessed the use of hsTnT assays, and the mean sensitivity of these tests in the diag nosis of acute MI was found to be 0.94 [95% confidence interval (CI) 0.89-0.97], and the mean sensitivity of con ventional tests was 0.72 (95% CI 0.63-0.79) but the mean specificity was 0.73 (95% CI 0.64-0.81) for the high sen sitivity assays compared with 0.95 (95% CI 0.93-0.97) for the conventional assays, and these differences in the   also associated with cardiovascular risk, heart failure and mortality as independent from conventional risk markers and hsTnT in patients with stable CAD [19]. Correia et al. showed that high sensitivity troponin I predicts cardiovas cular events similarly to conventional troponin T in the setting of nonSTelevation ACS [20]. However, Ndrepepa et al. pointed out that the elevated levels of hsTnT in patients with stable or unstable angina presenting with undetectable cTnT are significantly associated with reduced survival [21]. They also took attention that the use of hsTnT instead of cTnT significantly improved risk strati fication regarding longterm mortality but increased the proportion of patients with NSTEMI among patients with non ST elevation acute coronary syndrome [22]. Although the elevation of high sensitive troponin levels predicts higher risk of mortality in patients with sus pected ACS [16,[21][22][23]. The lower specificity of highsensi tivity assays results in higher rates of false positive tests, so some patients are incorrectly considered as undergo ing NSTEMI, and the use of high sensitive troponin assays leads to performing additional investigations, more coro nary angiography and interventions [24,25]. These addi tional investigations and interventions may pose the risk of the increased work load on emergency departments, cardiology referrals and cardiac catheterization laborato ries, and bring additional costs to the health care system, and also cause increased anxiety to patients [25].
In other causes of cardiac injury including pulmo nary embolism, myocarditis, pericarditis, congestive heart failure, septic shock, myocardial contusion, eleva tion of troponins has been shown to be associated with a bad outcome in most cases [26]. In cardiac contusion, the role of troponin measurement has been debated. The right ventricle is most commonly injured in blunt trauma, but the tissue volume of the right ventricle is reduced, which leads to concern that troponin levels will not ade quately elevate with blunt myocardial contusion. Alone, an elevated troponin is inadequate to diagnose myocar dial contusion but when it is combined with an abnormal electrocardiogram, patients are at risk for increased com plications requiring intervention, but when a negative 6h troponin is combined with a normal electrocardiogram and a hemodynamically stable patient, the negative pre dictive value for significant blunt myocardial contusion approaches 100% [27]. To date, the role of high sensitive troponin and conventional troponin assays in these cir cumstances has not been revealed well.
The present study does have several limitations. This was a single center study. The sample size in our study was relatively small. Long term follow up data were not available to reveal the clinical and prognostic importance of the results. The diagnosis of AMI was performed via coronary angiography and echocardiography. To use the tools showing the loss of myocardial viability such as positron emission tomography and thallium scintigraphy would have been better. Although it has recently been reported that triglyceride levels were higher in all age groups of AMI [28], in our study, triglyceride levels were significantly higher in AMI (-) group than AMI (+) one. This contrary result to current literature is probably due to play of chance. Few very high triglyceride levels may have caused this result. Despite all limitations, our study com paring the diagnostic power of four different troponin kits frequently used in the diagnosis of AMI provides impor tant contribution to the establishment of correct diagnosis of AMI.

Conclusions
The performance of a new developed Toyo Turklab Con ventional Troponin I (cTnI) assay was evaluated in this study in addition to three other troponin methods. There were no statistically significant differences between all four (Radiometer, Roche, Triage and Toyo Turklab) differ ent troponin assays in comparison of PPV and NPV in the diagnosis of AMI, but low sensitivity of Triage and Toyo (Turklab) assays should be considered.