NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

High-Sensitivity Cardiac Troponin for the Rapid Diagnosis of Acute Coronary Syndrome in the Emergency Department: A Clinical and Cost-Effectiveness Evaluation [Internet]. Ottawa (ON): Canadian Agency for Drugs and Technologies in Health; 2013 Mar. (CADTH Optimal Use Report, No. 2.1.)

Cover of High-Sensitivity Cardiac Troponin for the Rapid Diagnosis of Acute Coronary Syndrome in the Emergency Department: A Clinical and Cost-Effectiveness Evaluation

High-Sensitivity Cardiac Troponin for the Rapid Diagnosis of Acute Coronary Syndrome in the Emergency Department: A Clinical and Cost-Effectiveness Evaluation [Internet].

Show details

4CLINICAL REVIEW RESULTS

4.1. Selection of primary studies

A total of 1,163 potential citations were identified by the systematic search, with 1,046 citations being excluded during the title and abstract review based on irrelevance to the questions of interest. The full text documents of the remaining 118 articles were retrieved. Of these 118 articles, 96 did not meet the eligibility criteria and were excluded. Twenty-six articles were excluded because they were published only in an abstract form. Three of the excluded studies were duplicate or older publications of the already included studies. Twenty-two articles, reporting 16 studies, met the inclusion criteria and were included in this review. Figure 1 shows the PRISMA flowchart of the process used to identify and select studies for the review and the main reasons for exclusion. The list of the excluded studies and the reasons for exclusion are provided in Appendix 6.

Figure 1. Study Selection Flow Diagram — Clinical Review.

Figure 1

Study Selection Flow Diagram — Clinical Review.

4.2. Study characteristics

An overview of the included studies is provided in Table 1. All of the included articles were published between 2009 and 2012, with the majority published in 2011 (54%). Four of the 16 included studies were conducted in Germany,1417 three in New Zealand,1820 three in the United States,2123 two in Canada,24, 25 one in the United Kingdom,26 one in France,27 and one in Sweden.28 The APACE study was conducted in multiple countries in Europe.29 The patients enrolled in all studies presented with chest pain to EDs or chest pain units. All studies enrolled adult patients. However, the inclusion criteria were limited to patients older than 21 years of age in three studies,21, 22, 28 older than 25 years of age in one study,26 and between the ages of 18 and 85 in one study.16

Table 1. Characteristics of the included Diagnostic Studies.

Table 1

Characteristics of the included Diagnostic Studies.

Fifteen of the included studies were diagnostic studies with a single cohort design, that is, all patients received all of the study tests and the reference standard, with the primary outcome of the study being diagnostic performance of cTn tests. One study25 retrospectively selected a cohort of patients who had both hs-cTnI and AccuTnI tests performed using their “earliest presentation” blood samples. This study compared various concentrations of each test to the lowest concentration group in the same test result category, to evaluate the ability of the tests to predict death and MI. In all but one28 of the diagnostic studies reference standard was uniformly reported to be the final diagnosis of AMI or ACS based on available clinical, laboratory and imaging information. Lindahl et al. (2010)28 did not report the method for adjudicating of the final diagnosis, although they seem to have used the final diagnosis of AMI as the reference standard for evaluation of diagnostic accuracy of cTn tests. The final diagnoses were reported to be determined by two cardiologists (10 studies),1421, 27, 29 one cardiologist and one emergency physician (one study),24 two unspecified physicians (one study),23 or two investigators with MD and PhD degrees (one study).26 Two studies did not report on the number and specialty of the adjudicator(s).22, 28 The diagnosis of AMI was based on the universal definition of MI provided by the joint task force of the ESC, ACC, AHA, and the World Heart Federation task force (2007) in four studies,14, 15, 17, 27 and the ESC/ACC 2000 definition of MI in three studies.1820 The remaining studies did not specify the definition they used for the diagnosis of AMI.

All of the studies consistently administered both high-sensitivity and cTn tests to all participants at presentation to ED. The details of the reference standard and cTn tests used, as well as times of measurement are provided in Table 1.

4.3. Critical Appraisal of Individual Studies

Table 2 summarizes the results of the QUADAS-2 assessments. Six studies were rated as being in a high risk of selection bias due to a failure to recruit a consecutive or random sample of patients,15, 1921, 23, 25 four of those restricted their recruitment period to the morning to evening hours and excluded the patients referring to ED at night.1921, 23 The likelihood of selection bias due to convenience sampling was discussed in one of these studies.19 However, the authors reported they had missed only a small proportion of eligible patients who presented outside the recruitment hours. The authors of the three other studies did not discuss the restricted recruitment time as a limitation of their study.20, 21, 23 One study included patients with a negative cTnT at baseline,15 and one included patients for whom the results of cTn tests, performed on retrospectively collected blood samples, were available.25 In six studies16, 17, 22, 26, 28, 29 it was unclear whether their exclusion criteria could have introduced any selection bias, four of them excluded pregnant patients or those with concurrent non-cardiac conditions that might have affected their cTn levels, such as renal failure or active malignancy;16, 22, 26, 29 and in two studies patients were recruited in a retrospective manner.17, 28

Table 2. Risk of Bias and Applicability in the included Diagnosis Studies (results of QUADAS-2 quality assessment).

Table 2

Risk of Bias and Applicability in the included Diagnosis Studies (results of QUADAS-2 quality assessment).

Two studies reported blinding of the results of both high-sensitivity and cTn tests at the time of determining the final diagnosis.18, 24 In the remaining studies, the results of the cTn test (one of the index tests of interest in this review) were used to establish the diagnosis of MI or ACS (the reference standard).These studies were rated as having an unclear risk of bias in the index test domain.

None of the studies reported any specific information to assess whether the investigators were blinded to the reference standard results (final diagnosis) at the time of interpreting the results of cTn tests. This might raise a theoretical bias concern, especially for the retrospective studies. However, because in all of the studies the clinical diagnosis of MI (as the reference standard) required multiple clinical examinations and diagnostic tests, the results of which would have routinely been reported to the investigators who interpreted the results of index (cTn) tests, we decided it was impossible to blind the diagnosis process. Therefore, the risk of bias for reference standard was scored as low in all of the included studies. In addition, it was not possible to answer the question about the “appropriate interval between the index test and reference standard” because the exact time of confirmation of diagnosis (reference standard) was not reported by the authors of the included studies, due to the longitudinal nature of the clinical decision-making process.

Four studies were classified as being in high risk of bias for the flow and timing domain, in which a proportion of the recruited patients were excluded from the analysis due to technical problems, controversial management, insufficient serum sample, or missing test results.14, 15, 21, 26

Overall, applicability concerns related to all three domains were low for all of the included studies. In order to be rated as applicable to the research question, the studies should have included the same patient population, same index tests (cTn assays), and same reference standard as were defined in our study questions. One study, which exclusively included patients with an objective sign of cardiac ischemia, was classified as raising higher levels of applicability concern in terms of patient selection.

4.4. Data analyses and synthesis

4.4.1. Diagnostic test performance of hs-cTnT and hs-cTnI assays compared with each other as well as with conventional cTnT and sensitive cTnI assays in patients with suspected ACS symptoms in the ED?

a. Diagnostic performance of cTn assays compared with the reference standard (direct comparisons)

Diagnostic performance of a single cTn test administered at ED presentation

Overall seven studies contributed to the pooled analysis of the diagnostic accuracy of a single cTn test sample for diagnosis of AMI at the time of ED presentation.15,16,18,19,21,26,29 Different cTn tests were considered as index tests and final diagnosis of AMI served as the reference standard in our analysis. To avoid heterogeneous data, this analysis excluded the studies if they reported on specific subgroups of patients, excluded STEMI patients from their analyses, or reported the diagnosis accuracy of the tests performed in time points other than ED presentation.

In the selected studies, cTn assays were used with different diagnostic threshold levels: limit of detection (LoD), that is the lowest concentration of cTn that can be reliably detected by a testing procedure; 99th percentile cut-off point, that is the 99th percentile of the values for a reference control group (healthy population); and 10% coefficient of variation (CV), that is the lowest concentration with an acceptable imprecision (CV < 10%). The cut-off values were different for each type of cTn test and varied from one study to another (Table 1). We only included accuracy data related to cut-off points that had been determined a priori by the investigators and used as a part of laboratory diagnostic criteria. The reported information related to the cut-off points determined in a data-dependent manner, based on ROC curve analyses, was excluded from our pooled analyses. ROC-derived thresholds maximize both sensitivity and specificity and their inclusion in a meta-analysis might result in overestimation of actual diagnostic accuracy.

Data on the performance of a single hs-cTnI, hs-cTnT, cTnI, or cTnT, carried out at ED presentation for diagnosis of AMI, were available from three,16,21,22 five,15,18,19,26,28 six,16,18,19,21,26,29 and four18,22,26,29 studies. Reichlin et al.29 used three different cTnI assays in a single study. Therefore, data from multiple arms of this study were included in the pooled analyses related to cTnI. Table 3 presents the sensitivity and specificity measures of the cTn assays reported in these studies. The results of direct pooled analyses of cTn test versus the gold standard (final diagnosis of AMI) at the time of presentation are provided in Table 4. As it is shown in this table, among the four types of cTn assays, the highest sensitivity of a single sample for diagnosis of AMI was found for hs-cTnI at LoD threshold (1.00, 95% CI, [0.98 to 1.00]; one study;16 n = 1,818) and the highest specificity was for cTnT at 10% CV threshold (0.97 to 95% CI, [0.96 to 0.98]; two studies;18,29 n = 1,050). The pooled analysis also showed that the summary estimates of sensitivity for all four types of cTn assays were consistently higher at the cut-off point of LoD; whereas, no similar pattern was detected for summary estimates specificity. Data from the study by Bhardwaj et al. (2011; n = 318)22 in which the diagnostic accuracy of hs-cTnI for ACS were compared with that of cTnT was not included in the pooled analysis of baseline measurement data because this study reported to have performed cTn tests of interest at two to four hours after ED admission. In this study, the sensitivity and specificity of hs-cTnI was reported to be 0.57 (95% CI, 0.45 to 0.67) and 0.86 (95% CI, 0.81 to 0.90) respectively. The sensitivity and specificity values for cTnT were 0.22 (95% CI, 0.14 to 0.34) and 0.97 (95% CI, 0.94 to 0.99) respectively, at two to four hours after ED admission.

Table 3. Pooled Analysis of Sensitivity and Specificity of Troponin Tests for Diagnosis of AMI, at the Time of ED Presentation.

Table 3

Pooled Analysis of Sensitivity and Specificity of Troponin Tests for Diagnosis of AMI, at the Time of ED Presentation.

Table 4. Pooled Accuracy Estimates of Troponin Tests in Diagnosis of AMI, When Administered at the Time of ED Presentation.

Table 4

Pooled Accuracy Estimates of Troponin Tests in Diagnosis of AMI, When Administered at the Time of ED Presentation.

b. Comparison of diagnostic thresholds of cTn assays (single measurement)

The pooled estimates of sensitivity and specificity for different cut-off points of the four types of cTn tests are also shown in summary ROC curves (Figures 2 to 5).

Due to paucity of studies utilizing hs-cTnI, we could not compute the diagnostic performance of this test at the 10% CV cut-off point. The sensitivity of hs-cTnI at LoD (1.00; 95% CI, [0.98 to 1.00], one study;16 n = 1,818) was statistically higher than that of the 99th percentile cut-off point (0.82; 95% CI, [0.79 to 0.85], two studies;16,21 n = 2,204) and its specificity (0.31; 95% CI, [0.28 to 0.34], one study;16 n = 1,818) was statistically lower than that of the 99th percentile cut-off point (0.824; 95% CI, [0.790 to 0.854], two studies;16,21 n = 2,204) (Table 4 and Figure 2).

Figure 2. Summary of ROC Curve for hs-cTnI at ED Presentation.

Figure 2

Summary of ROC Curve for hs-cTnI at ED Presentation.

There was no statistically significant difference between sensitivities of hs-cTnT at 99th percentile and 10% CV cut-off points. However, the sensitivity of this test at LoD (0.97; 95% CI, [0.96 to 0.98], three studies;18,26,29 n = 1,753) was statistically higher, and its specificity at this threshold (0.326; 95% CI, [0.321 to 0.329], three studies;18,26,29 n = 1,753) was statistically lower than those of the two other cut-off points (Table 4 and Figure 3).

Figure 3. Summary of ROC Curve for hs-cTnT at ED Presentation.

Figure 3

Summary of ROC Curve for hs-cTnT at ED Presentation.

The sensitivities and specificities of all three cut-off points of cTnI were statistically different, with LoD being the most sensitive (0.92; 95% CI, [0.90 to 0.93], three studies;16,18,29; n,= 2,868) and least specific (0.808; 95% CI, [0.804 to 0.812], three studies;16,18,29; n,= 2,868) and the 99th percentile threshold being the most specific (0.94, 95% CI, [0.92 to 0.96], five studies;16,18,19,21,29 n,=,3,712) and least sensitive (0.81, 95% CI, [0.76 to 0.85], five studies;16,18,19,21,29 n,=,3,712) among the three cut-off points (Table 4 and Figure 4).

Figure 4. Summary of ROC Curve for cTnI at ED Presentation.

Figure 4

Summary of ROC Curve for cTnI at ED Presentation.

The sensitivity of cTnT at 10% CV threshold (0.48, 95% [CI, 0.46 to 0.50], two studies;18,29 n = 1,050) was statistically lower and its specificity (0.97, 95% CI, [0.96 to 0.98], two studies;18,29 n = 1,050) was statistically higher than the two other cut-off points for this test (Table 4 and Figure 5). No statistically significant differences were found between the 99th percentile threshold and LoD for cTnI in terms of sensitivity and specificity.

Figure 5. Summary of ROC Curve for cTnT at ED Presentation.

Figure 5

Summary of ROC Curve for cTnT at ED Presentation.

When ROC curves were constructed for different cut-off points of each cTn assay, the AUC for cTn tests were not statistically different, based on the overlapping CIs.

c. Diagnostic performance of serial cTn tests

Due to the paucity of data on serial measurements of the cTn assays no pooled analysis was conducted to evaluate the diagnostic accuracy of combinations of consecutive cTn tests. Overall, three of the included studies reported on the performance of serial cTn assays.14,16,27 Two of these studies reported the diagnostic accuracy of serial testing in diagnosis of AMI;14,27 whereas, one study reported the accuracy of relative changes of cTn concentration during consecutive measurements.16 All three studies compared hs-cTnT with cTnI.

The diagnostic accuracy measures of serial cTn testing from the available studies are summarized in Table 5. As the Table shows, the sensitivity and specificity of serial hs-cTnT ranged from 0.9327 to 0.9814 and from 0.4114 to 0.8227 respectively. The sensitivity of serial cTnI varied between 0.7127 and 0.9114 and its specificity between from 0.4614 to 0.97.27

Table 5. Reported Diagnostic Accuracy of Serial CTn Testing in Diagnosis of AMI.

Table 5

Reported Diagnostic Accuracy of Serial CTn Testing in Diagnosis of AMI.

d. Subgroup analyses

Data on the following patient subgroups were available from the included studies. These subgroups were not pre-specified. In all of these studies different cTn (index) tests were compared with the final diagnosis of AMI (as the reference standard).

NSTEMI

Aldous et al. (2011)20 focused their analysis on NSTEMI patient (n = 1,000), hs-cTnT was compared with point-of-care cTnI (POC-cTnI).The sensitivity and specificity of hs-cTnT for diagnosis of AMI at a cut-off point of 99th percentile were found to be 0.91 (95% CI, 0.87 to 0.94) and 0.81 (95% CI, 0.80 to 0.82) respectively. POC-cTnI showed significantly lower sensitivity (0.62, 95% CI, 0.58 to 0.66), and higher specificity (0.96; 95% CI, 0.94 to 0.97) values.

Timing of assessment

Subgroup data on timing of assessment were available for two studies.26,29 Table 6 summarizes the diagnostic performance measures of cTn tests in the subgroup of patients with early versus late presentation. Both studies considered a three-hour period after the onset of symptoms as a cut-off point for definition of early presentation. Body et al.26 used an additional six-hour cut-off point to distinguish between early and late presentation. As the table shows, hs-cTnT showed sensitivity and specificity values higher than 80% when it was used at a 99th percentile cut-off point, regardless of timing of assessment. The point estimates of sensitivity were perfect (1.00) in both studies when using LoD cut-off points, but specificity values were lower and variable at this threshold. By contrast, cTnT had much lower levels of sensitivity (0.4429 to 0.6626 in patients who presented within three hours and 0.7126 in those who presented within six hours from the onset of chest pain), compared with those in the late presentation subgroup where the sensitivity of cTnT improved considerably (0.92 in patients who presented within three hours and 0.93 who presented within six hours from the onset of chest pain).26 The specificity estimates for cTnT maintained between 0.82 and 0.93 in the late presentation subgroup.26

Table 6. Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on the Time Period Between the Onset of Symptoms and ED Presentation.

Table 6

Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on the Time Period Between the Onset of Symptoms and ED Presentation.

Reichlin et al. (2009)29 reported the diagnostic performance of cTnI in patients who presented within three hours of their symptoms (Table 6) but did not report the results of late measurements. Based on the results of this study, the sensitivity values of cTnI assays for detection of AMI were maximized (≥ 0.85), when they were used at LoD thresholds. The specificity values were relatively stable across different cut-off points, except for Siemens-Ultra cTnI, which yielded a significantly lower specificity when using LoD (0.74; 95% CI, [0.67 to 0.80], as compared with the 99th percentile cut-off point (0.95; 95% CI, [0.90 to 0.97]).

Baseline risk stratification

One of the included studies27 reported the diagnostic performance of cTn tests in subgroups of patients with high risk of MI at baseline (n = 59) and those with intermediate or low risk of MI at baseline (n = 258). One additional study23 limited its patient population to low and intermediate-risk patients (n = 377). The findings of these studies are shown in Table 7. Januzzi et al. (2010)23 reported similar sensitivity values for hs-cTnT and cTnT in low and moderate-risk patients (0.88, 95% CI, [0.47 to 1.00]). However, the comparison of the reported estimates of specificity and their CIs shows that the specificity of cTnT (0.94 with 95% CI [0.92, 0.97] and 0.97 with 95% CI, [0.95 to 0.98] for the 99th percentile and LoD thresholds respectively) was statistically higher than that of hs-cTnT (0.85; 95% CI, [0.81 to 0.89]) in patients with a low-to-moderate risk of disease.

Table 7. Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on the Pretest Probability of MI.

Table 7

Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on the Pretest Probability of MI.

In the study by Freund et al. (2011),27 hs-cTnT was shown to have similar sensitivity values in high and low-risk subgroups, but the specificity of the test was statistically higher in the low-to-moderate-risk population (0.85 [95% CI, 0.79 to 0.89] versus 0.67 [95% CI, 0.49 to 0.81] in high-risk group). When compared with hs-cTnT, in both risk categories, cTnI was reported to have relatively lower sensitivity (0.77 [95% CI 0.54 to 0.92] versus 0.91.(95% CI, 0.69 to 0.98) in low risk and 0.65 [95% CI, 0.43 to 0.83] versus 0.96 [95% CI, 0.76 to 1.00] in high-risk populations), but higher specificity values (0.97 [95% CI, 0.94 to 0.99] versus 0.85 [95% CI, 0.79 to 0.89] in the low-risk group and 0.94 [95% CI, 0.79 to 0.99] versus 0.67 [95% CI, 0.49 to 0.81] in the high-risk populations) (Table 7).

History of ischemic heart disease

One study reported on the diagnostic performance of hs-cTnT, cTnT, and two cTnI (Abbott Architect and Siemens-Ultra) assays in diagnosis of AMI, based on the patients’ history of ischemic heart disease (IHD).30 As it is demonstrated in Table 8, in both subgroups of patients with (n = 401) and without (n = 697) a history of IHD, hs-cTnT was statistically more sensitive than cTnT, according to non-overlapping CIs. However, there was no statistically significant difference between hs-cTnT and cTnI assays. When compared between the two subgroups of patients with and without a history of IHD, hs-cTnT had similar sensitivity values. However, it was statistically more specific in patients with a negative history (0.81 [95% CI, 0.78 to 0.84] versus 0.59 [95% CI, 0.54 to 0.65] in patients with a positive history). In both subgroups, cTnI assays were reported to have sensitivity and specificity values that were greater than 80%. The specificity of these assays was slightly higher in patients with no previous ischemic heart disorders.

Table 8. Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on Previous History of IHD in Patients with Chest Pain Who Are Referred to ED.

Table 8

Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on Previous History of IHD in Patients with Chest Pain Who Are Referred to ED.

Baseline cTn test results

The diagnostic performance of repeated cTn assays was evaluated in three studies,16,33,34 based on the baseline cTn test results. An overview of the diagnostic accuracy estimates from these studies is presented in Table 9. One of the three studies, that excluded STEMI patients from the analysis, reported the accuracy of cTn tests in the diagnosis of NSTEMI from UA only in patients who had initially positive test results.34 The two other studies that had AMI as the outcome of interest reported on the diagnostic performance of cTn tests in patients with a positive or negative test at baseline.16,33 All three studies compared hs-cTnT to a cTnI assay.

Table 9. Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on the Results of Baseline Troponin Test.

Table 9

Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on the Results of Baseline Troponin Test.

Comparison of the sensitivity and specificity values estimated from data presented by Reichlin et al. (2011)33 showed that, in patients with a positive baseline test, hs-cTnT was statistically more sensitive and less specific than Siemens cTnI-ultra, when these cTn tests were performed two hours after an initially positive cTn assay (Table 9). In patients who had negative test results at baseline; however, this study found no statistically significant difference between hs-cTnT and cTnI-Ultra assays, which were performed one or two hours after ED admission. The differences reported by the two other studies16,34 were also non-significant in either of the subgroups.

Age groups

Reiter et al. (2011)31 reported on the diagnostic performance of hs-cTnT, cTnT, and two cTnI (Abbott Architect and Siemens-Ultra) assays in elderly patients (> 70 years of age; n = 406) and those who were under 70 years of age (n = 681) (Table 10). This study found no significant difference between hs-cTnT and the cTnI assays in terms of diagnostic performance in either of the subgroups. However, as it is shown in Table 10, the sensitivity values of cTnT in both subgroups (0.59 [95% CI, 0.47 to 0.70; 10% CV threshold] for ≤ 70 years, and 0.83 [95% CI, 0.84 to 0.90; LoD threshold] or 0.76 [95% CI, 0.57 to 0.79; 10% threshold] for > 70 years age groups) were statistically lower than those of hs-cTnT (0.88 [95% CI, 0.78 to 0.94; 99th percentile threshold] for under 70, and 1.00 [95% CI, 0.96. 1.00; LoD threshold] or 0.98 [95% CI, 0.93 to 1.00; 99th percentile threshold] for 70 plus age groups). On the contrary, the specificity values of cTnT (0.98 [95% CI, 0.97 to 0.99; 10% CV threshold] for under 70, and 0.90 [95% CI, 0.86 to 0.93; LoD threshold] or 0.96 [95% CI, 0.93 to 0.98; 10% threshold] for > 70 years age groups) were statistically higher than those of hs-cTnT (0.86 [95% CI, 0.83 to 0.89; 99th percentile threshold] for under 70, and 0.01 [95% CI 0.00. 0.03; LoD threshold] or 0.49 [95% CI 0.0.44, 0.55; 99th percentile threshold] for > 70 age groups). Based on the construction of ROC curves for different diagnostic thresholds, the authors suggested that the optimum cut-off points were significantly higher in the elderly population than in younger patients.

Table 10. Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on Age Categories of Patients with Chest Pain Referred to ED.

Table 10

Reported Diagnostic Accuracy of Troponin Tests in Diagnosis of AMI, Based on Age Categories of Patients with Chest Pain Referred to ED.

e. Relative diagnostic accuracy of cTn assays (indirect comparisons)

Based on the 99th percentile cut-off point at ED presentation, Table 11 presents estimates of relative diagnostic accuracy for the diagnosis of AMI with the four cTn assays, using indirect comparisons. Overall diagnostic accuracy is represented by the AUC of the ROC curve. The relative sensitivity and specificity are also presented.

Table 11. Relative Diagnostic Performance of Troponin Tests for Diagnosis of AMI, When Used at a 99th Percentile Cut-off Point at the Time of ED Presentation.

Table 11

Relative Diagnostic Performance of Troponin Tests for Diagnosis of AMI, When Used at a 99th Percentile Cut-off Point at the Time of ED Presentation.

As shown in the table, overall relative diagnostic accuracy for hs-cTnT was statistically lower compared with both cTnT (0.942; 95% CI, 0.907 to 0.977), and cTnI (0.923, 95% CI, 0.886 to 0.961). Overall relative diagnostic accuracy for hs-cTnI is not statistically different compared with either cTnT (1.016; 95% CI, 0.993 to 1.040) or cTnI (0.996; 95% CI, 0.968 to 1.025). Comparing the high-sensitivity tests to each other finds hs-cTnI to have higher overall diagnostic accuracy compared with hs-cTnT (1.079; 95% CI, 1.038 to 1.122). Relative sensitivity for hs-cTnT was statistically higher compared with both cTnT (1.353; 95% CI, 1.272 to 1.439), and cTnI (1.089; 95% CI, 1.077 to 1.180). Relative sensitivity for hs-cTnI was found to be statistically higher compared with cTnT (1.270; 95% CI, 1.215 to 1.327), but not compared with cTnI (1.024; 95% CI, 0.949 to 1.104). Although hs-cTnT had higher relative sensitivity compared with hs-cTnI, it was not statistically significant (1.066; 95% CI, 0.998 to 1.138). Relative specificity for hs-cTnT was statistically lower compared with both cTnT (0.864; 95% CI, 0.844 to 0.886), and cTnI (0.875; 95% CI, 0.849 to 0.901). Relative specificity for hs-cTnI was found to be statistically lower compared with cTnT (0.942; 95% CI, 0.931 to 0.954) and to cTnI (0.953; 95% CI, 0.933 to 0.973). When comparing the two high-sensitivity cTn tests with each other, hs-cTnI had higher relative specificity compared with hs-cTnT (1.090; 95% CI, 1.064 to 1.116).

Based on the 10% CV cut-off point at ED presentation, Table 12 presents estimates of relative diagnostic accuracy for diagnosis of AMI with the four cTn assays, using indirect comparisons. As shown, overall relative diagnostic accuracy for hs-cTnT was statistically lower compared with cTnI (0.946; 95% CI, 0.926 to 0.967), but not compared with cTnT (0.983, 95% CI, 0.964 to 1.007). Relative sensitivity for hs-cTnT was statistically higher compared with cTnT (0.577; 95% CI, 0.534 to 0.623), but not cTnI (0.963; 95% CI 0.945 to 0.982). Relative specificity for hs-cTnT was statistically lower compared with cTnT (0.853; 95% CI, 0.824 to 0.886) and cTnI (0.905; 95% CI, 0.874 to 0.937). No data were available to compare diagnostic accuracy of hs-cTnI compared with other cTn tests.

Table 12. Relative Diagnostic Performance of Troponin Tests for Diagnosis of AMI, When Used at a 10% CV Cut-off Point at the Time of ED Presentation.

Table 12

Relative Diagnostic Performance of Troponin Tests for Diagnosis of AMI, When Used at a 10% CV Cut-off Point at the Time of ED Presentation.

Table 13 presents similar results using a cut-off point based on the LoD. Overall relative accuracy of hs-cTnT for diagnosis of AMI was statistically lower compared with cTnT (0.956; 95% CI, 0.934 to 0.978), and cTnI (0.940; 95% CI, 0.922 to 0.959). Relative sensitivity for hs-cTnT was statistically higher compared with both cTnT (1.469; 95% CI, 1.420 to 1.519), and cTnI (1.060, 95% CI, 1.040 to 1.080). Relative sensitivity for hs-cTnI was found to be statistically higher compared with cTnT (1.508; 95% CI, 1.459 to 1.559), but not cTnI (1.088; 95% CI 1.060 to 1.104). The relative sensitivity was higher for hs-cTnI compared with hs-cTnT (1.027; 95% CI, 1.011 to 1.042).

Table 13. Relative Diagnostic Performance of Troponin Tests for Diagnosis of AMI, When Used at a LoD Cut-off Point at the Time of ED Presentation.

Table 13

Relative Diagnostic Performance of Troponin Tests for Diagnosis of AMI, When Used at a LoD Cut-off Point at the Time of ED Presentation.

Relative specificity for hs-cTnT was statistically lower compared with both cTnT (0.344; 95% CI, 0.339 to 0.350) and cTnI (0.403; 95% CI, 0.398 to 0.409). Relative specificity for hs-cTnI was found to be statistically lower compared with cTnT (0.325; 95% CI, 0.294 to 0.359), but not compared with cTnI (0.381; 95% CI, 0.345 to 0.421). There was no statistically significant difference in the relative specificity of hs-cTnT compared with hs-cTnI (1.058; 95% CI, 0.958 to 1.170).

4.4.2. What is the clinical effectiveness of hs-cTnT and hs-cTnI assays compared with each other as well as with conventional cTnT and sensitive cTnI assays in patients with suspected ACS symptoms in the ED?

a. Major cardiovascular events

Mortality data were available from five studies.14,17,28,32,35 The incidence data reported by two studies14,28 are presented in Table 14. Estimates of hazard ratio for death and/or MI, using Cox proportional hazard analyses, and the accuracy of cTn tests in prognosis of cardiac or all-cause mortality (with or without MI) were reported in five14,25,28,32,35 and two32,35 studies respectively (Tables 15 and 16).

Table 14. Death and Incident AMI in Patients Diagnosed by cTn Tests Reported by the Included Studies.

Table 14

Death and Incident AMI in Patients Diagnosed by cTn Tests Reported by the Included Studies.

Table 15. Hazard Ratios of Major Cardiac Events and Death in Patients Diagnosed by cTn Tests.

Table 15

Hazard Ratios of Major Cardiac Events and Death in Patients Diagnosed by cTn Tests.

Table 16. Estimates of Diagnostic Accuracy of Troponin Tests (administered at ED presentation) for Major Cardiac Events and Death Reported in the Included Studies.

Table 16

Estimates of Diagnostic Accuracy of Troponin Tests (administered at ED presentation) for Major Cardiac Events and Death Reported in the Included Studies.

As it is shown in Table 14, the relative frequencies of incident MI and death during the study follow-up times were statistically higher in patients with a positive hs-cTnT than those with a positive cTnI or cTnT (except for the incidence of non-fatal MI that was not statistically different in patients with positive and negative cTnI tests).

Hazard ratios provided in Table 15 show that positive results of both high-sensitivity and cTn tests were associated with statistically higher risks of death or composite outcome of death or MI, as compared with those who had negative test results. However, the hazard ratios were two to three times higher for hs-cTnT than for cTnI or cTnT assays, suggesting that a positive hs-cTnT was a better independent prognostic factor of mortality. Christ et al.17 found that patients with dynamic changes of 30% or more had the highest risk for death or MI (P < 0.001). The results of their study are not shown in the table, since the authors did not report any details of hazard data in their publication.

Based on the data published by Mueller at al. (2012; n = 1,384),14 the frequency of patients who died after they were discharged with a negative hs-cTnT (one death) was five times smaller than the number of deaths in those who had a normal cTnI (five deaths). Similarly, the number of patients with a subsequent MI was 2.4 times lower in patients who were discharged with a negative hs-cTnT (five non-fatal MIs), compared with those who had a negative cTnI test (12 non-fatal MIs). The authors suggested that the hazard of discharging patients with normal values of cTn tests is “substantially’ lower when hs-cTnT is used, as compared with discharging based on a normal cTnI test. In the study by Hochholzer et al. (2011),32 after adjustment for the baseline MI risk score, hs-cTnT was not a prognostic factor for incident MI (P = 0.23) or the composite outcome of cardiac death or non-fatal MI (P = 0.06) (Table 15). This study found hs-cTnT to be statistically more sensitive, but less specific than cTnT in prognosis of death alone (Table 16). Kavsak et al.(2009; n = 383)25 compared patient subgroups who had various cTn concentrations measured by hs-cTnI or AccuTnI to those with lowest concentrations of the same cTn test, in terms of short and long-term risk of death and/or MI. As shown in Table 15, the results of their Cox proportional hazard analysis showed that patients with hs-cTnI or AccuTnI values greater than 0.04 mcg/L were in a statistically higher risk of 30-day MI and the composite outcome of death or MI at any study time point from 30 days to 10 years, when compared with patients in the lowest concentration group of the same cTn test. Neither hs-cTnI nor AccuTnI could statistically predict short-term mortality (30 days). However, an hs-cTnI value greater than 0.01 mcg/L was shown to be an independent prognostic factor for mortality at one (P = 0.04) or two years (P < 0.01). The corresponding hazard ratios for AccuTnI were not statistically significant for the prediction of death at one or two years.

The prognostic accuracy of cTn test for composite outcomes that included major cardiac events outcomes was reported in three studies24,25,35 (Table 16). Kavsak et al. (2012; n = 186)24 defined “serious cardiac outcomes” as the composite of MI, heart failure, and arrhythmias. They reported fairly similar AUC of the ROC curve estimates for hs-cTnT, hs-cTnI, and cTnI assays. In a different study by the same author25 AccuTnI achieved a statistically greater AUC of the ROC curve (0.81, 95% CI, [0.73 to 0.90]) for the composite outcome of death or MI at 30 days, as compared with the AUC for hs-cTnI (0.74, 95% CI, (0.66 to 0.82); P = 0.05) Aldous et al. (2011)35 included cardiac death, non-fatal MI, and revascularization in their composite outcome named MACE (major adverse cardiovascular events). In this study, the AUC of the ROC curves for hs-cTnT, cTnT, and cTnI tests were statistically predictive of two-year cumulative MACE that was significantly higher for hs-cTnT (AUC = 0.70, 95% CI, [0.63 to 0.76]) than for cTnT [AUC = 0.61, 95% CI, [0.63 to 0.76]; P = 0.001). However, there was no statistical difference between hs-cTnT and cTnI (AUC = 0.86, 95% CI, [0.78 to 0.95]; P = 0.094) in terms of diagnostic accuracy for two-year MACE. In addition, this study found no statistical differences between hs-cTnT, cTnT, and cTnI in terms of subsequent revascularization procedures, based on overlapping CIs for estimates of AUC of the ROC curve.

b. Quality of life

The review found no evidence reporting on the effects of cTn tests on quality of life outcomes.

c. Readmission rates

Readmission rates were not reported in any of the included studies.

d. ED time until diagnosis or detection of abnormal concentration

No description related to ED times between the performance of cTn tests and the diagnosis of MI or ACS was found in the included studies.

Copyright © 2013 Canadian Agency for Drugs and Technologies in Health.
Bookshelf ID: NBK168941
PubReader format: click here to try

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.5M)

Other titles in this collection

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...