Variability of High-Sensitivity Troponin T Concentrations in Emergency Settings: Impact for the Diagnosis of Myocardial Infarction

Variability of High-Sensitivity Troponin T Concentrations in Emergency Settings: Impact for the... Abstract Objectives To assess biological variation of troponin T in emergency settings and establish limits for interpretation of serial results. Methods We studied 6,557 consecutive patients with troponin measurements. A stable reference subset was selected to estimate biological variation and threshold limits. Results The first troponin level was elevated in 32% of patients, and 2,490 had a second troponin level with a myocardial infarction (MI) prevalence of 16.2%. In the stable reference group with at least one abnormal value, the 99th percentile of the absolute delta between the first two samples was 16 ng/L. For MI diagnosis, the area under the receiver operating characteristic curve was 0.85 (confidence interval [CI], 0.83-0.87) for the first troponin level and 0.94 (CI, 0.93-0.95) for the absolute delta. Conclusions An absolute delta of 16 ng/L has good specificity in the emergency setting. This threshold is valid for any sex, age, and sampling interval between 3 and 24 hours and is higher than published limits found in healthy outpatients. Myocardial infarction, Troponin, Acute coronary syndrome, Biological variation, Reference range Many individuals without acute conditions have basal high-sensitivity troponin T (hs-TnT) values repeatedly above 14 ng/L, the recommended 99th percentile threshold for the diagnosis of myocardial infarction (MI). Factors associated with increased values include older age, male sex, chronic heart disease, peripheral atherosclerotic vascular disease, hypertension, diabetes, and chronic kidney disease.1-4 For patients with chronic elevations of troponin T, the challenge is to differentiate “stable” patients from patients with an acute condition that increases troponin, especially an MI. Some authors have suggested to use age and sex-adjusted upper limits to improve discrimination.2,5,6 However, dynamic changes are still required in most cases to identify MI with certainty.7 Moreover, patients presenting to the emergency room in a clinical context are more likely to have comorbidities and have a chronic elevation of hs-TnT than typically studied reference populations with young and healthy individuals. Our objectives were to assess biological variability of hs-TnT in unselected patients presenting to the emergency room and define a 99th percentile limit for the delta in sequential troponin results in stable patients without acute injury. This limit could then be applied as a rule-in criterion to identify acute cardiac injury. Secondary objectives were to evaluate the diagnostic accuracy of the criterion to recognize MI and to identify factors associated with increased biological variability. Materials and Methods We performed a retrospective study of consecutive patients presenting to the emergency department of a university hospital (IUCPQ, Québec, Canada). In total, 6,557 consecutive patients with at least one hs-TnT measurement were included. Only the first encounter was studied for each patient. hs-TnT was measured in heparinized plasma on a stat basis with modular analyzers (Roche Diagnostics, Montreal, Canada) according to the manufacturer’s instructions. Imprecision is below 10% for values above 10 ng/L. All results are expressed as whole numbers without decimals. We predefined the following cutoffs: (1) 3 ng/L as the limit of detection, with any value below 3 set at 3 ng/L for statistical analyses; (2) 14 ng/L as the upper limit of normal, as suggested by the manufacturer; and (3) 50 ng/L as equivalent to the upper reference limit used with the previous troponin T assay to identify a high-risk group. Variation of troponins was analyzed in patients with repeated troponin measurements within a limit of 24 hours (n = 2,490). Our strategy was to select a reference cohort from the whole cohort with a very low probability of acute events, apply the principle of symmetry as described below, determine a 99th percentile limit of variation, and evaluate its performance in the whole cohort. We focused our analysis on patients with normal or slightly abnormal (below 50 ng/L) levels since this is the critical zone where sequential analysis is most needed. Patients with MI or other acute events typically show increases in troponin. In contrast, patients without acute events may show either increasing or decreasing sequential values in a random manner. Patients who have no change (identical results) in troponin or decreasing values with an average below 50 ng/L are at very low risk of MI. McMullin et al8 found that a decreasing troponin T in an emergency setting was not associated with increased odds of acute coronary syndrome in a cohort of over 1,800 patients. Among patients with a mean hs-TnT less than 50 ng/L (n = 1,861), we thus defined a reference cohort of stable patients with equal or decreasing results. For stable patients, it is expected that hs-TnT values vary randomly around a mean such that for two sequential results, there are equal chances to have R1 > R2 or R2 > R1, with half going up and half going down. Thus, the absolute variation can be used as a surrogate of variability in stable low-risk patients (principle of symmetry). To avoid statistical bias, we included only half of the patients with identical values selected randomly and excluded extreme outliers defined as a change exceeding three times the interquartile range (a standard criterion for extreme outliers). Figure 1 describes the selection process for the reference cohort. Figure 1 View largeDownload slide Cohort selection flowchart. Figure 1 View largeDownload slide Cohort selection flowchart. We evaluated the diagnostic performance of the threshold found in the reference cohort study for the identification of MI. In the absence of an independent gold standard, we used the final diagnosis from the electronic medical records as established by the treating physicians without any patient chart reviews. Complementary investigations were at the discretion of the clinicians, and all troponin results were available to physicians for their final judgment. The hospital research ethics review board approved the study. The validation cohort included patients with at least two hs-TnT measurements within a 24-hour period (Figure 1; n = 2,490). Because of the uncertainty about troponin evolution, patients with a diagnosis of cardiomyositis, pericardial disease, or circulatory shock were excluded (n = 52). All others (n = 2,438) were included in the validation cohort and were divided into positive or negative for MI. Patients with cardiac arrest were included in the MI group (Figure 1). Statistical Analyses We made comparisons between groups with the standard t test, the χ2 test, and the nonparametric Kruskal-Wallis (KW) test or Mann-Whitney U test where appropriate. Outliers and extreme outliers were defined as, respectively, any value exceeding 1.5 and 3 times the interquartile range from the median. In the reference group, we defined the threshold to identify a significant change in hs-TnT based on the 99th percentile of the delta distribution between repeated measures. Variables associated with the absolute delta between two consecutive troponin results were analyzed by univariate and multivariate linear regression statistics in the reference group. To evaluate the performance of each testing strategy, we calculated sensitivity, specificity, and area under the receiver operating characteristic (ROC) curve. Logarithmic scale was used to better visualize results in the low range. Results We identified 6,557 consecutive patients with an hs-TnT measurement in the emergency department. Median age was 68 years, and 55% were men. There were 435 MI cases, representing 6.6% of patients. The first hs-TnT was normal (≤14 ng/L) in 68%, slightly elevated (between 15 and 50 ng/L) in 23%, and above 50 ng/L in 9%. In those three groups, MI was diagnosed respectively in 1.4%, 6.5%, and 46.6%. Of the whole cohort, 38% had a second sample within 24 hours (n = 2,490 with 16.2% MI). Of those, 1,861 had an average troponin below 50 ng/L, among which 445 had decreasing and 751 had identical values. Figure 1 and Table 1 describe the population and the subgroups studied. To select the reference stable group, we kept the patients with decreasing values (n = 445) and added half of the 751 patients with equal values selected randomly (n = 376), thus generating a set of 810 patients after exclusion of 11 extreme outliers (delta above 20 ng/L). Basic characteristics of these groups are displayed in Table 1. Differences between group B and the reference group are statistically significant for age, sex, and troponin distributions (P < .01), mainly due to the large sample size. However, we consider the reference group quite representative of the population served (Table 1). Table 1 Cohort Characteristics Groupa  Complete Cohort  Validation Cohort  Reference Cohort  No. of patients  6,557  2,438  810  Age, median, ya  68.3  70.4  68.8  Age, mean ± SD, ya  66 ± 16  69.4 ± 14  67.8 ± 14  Sex (male), %  55.4  58.9  55.2  First troponin T, median, ng/La  6  13  7   0-14 ng/L, %a  68  52.5  65.8   15-50 ng/L, %a  22.6  27.2  32.6   >50 ng/L, %a  9.4  20.3  1.6  Second troponin T, median, ng/La  —  16  5  Sampling interval, median, ha  —  6.7  6.2  Groupa  Complete Cohort  Validation Cohort  Reference Cohort  No. of patients  6,557  2,438  810  Age, median, ya  68.3  70.4  68.8  Age, mean ± SD, ya  66 ± 16  69.4 ± 14  67.8 ± 14  Sex (male), %  55.4  58.9  55.2  First troponin T, median, ng/La  6  13  7   0-14 ng/L, %a  68  52.5  65.8   15-50 ng/L, %a  22.6  27.2  32.6   >50 ng/L, %a  9.4  20.3  1.6  Second troponin T, median, ng/La  —  16  5  Sampling interval, median, ha  —  6.7  6.2  —, data not available because of absence of second sample in many cases. aComparison between cohorts is statistically significant (P < .01). View Large Table 1 Cohort Characteristics Groupa  Complete Cohort  Validation Cohort  Reference Cohort  No. of patients  6,557  2,438  810  Age, median, ya  68.3  70.4  68.8  Age, mean ± SD, ya  66 ± 16  69.4 ± 14  67.8 ± 14  Sex (male), %  55.4  58.9  55.2  First troponin T, median, ng/La  6  13  7   0-14 ng/L, %a  68  52.5  65.8   15-50 ng/L, %a  22.6  27.2  32.6   >50 ng/L, %a  9.4  20.3  1.6  Second troponin T, median, ng/La  —  16  5  Sampling interval, median, ha  —  6.7  6.2  Groupa  Complete Cohort  Validation Cohort  Reference Cohort  No. of patients  6,557  2,438  810  Age, median, ya  68.3  70.4  68.8  Age, mean ± SD, ya  66 ± 16  69.4 ± 14  67.8 ± 14  Sex (male), %  55.4  58.9  55.2  First troponin T, median, ng/La  6  13  7   0-14 ng/L, %a  68  52.5  65.8   15-50 ng/L, %a  22.6  27.2  32.6   >50 ng/L, %a  9.4  20.3  1.6  Second troponin T, median, ng/La  —  16  5  Sampling interval, median, ha  —  6.7  6.2  —, data not available because of absence of second sample in many cases. aComparison between cohorts is statistically significant (P < .01). View Large The absolute delta between the two measurements increased progressively (r = 0.56, P < .001) with increasing hs-TnT Figure 2A. However, for patients with values between 15 and 50 ng/L, the upper limit of dispersion was relatively stable, suggesting that a unique cutoff for absolute delta could be used. Relative change showed a progressive decrease in dispersion with increasing hs-TnT values Figure 2B. For the reference group of 810 patients, the 99th percentile of the absolute delta between samples 1 and 2 was 14 ng/L. However, for subjects with at least one value above 14 ng/L (a condition required for the diagnosis of MI), the calculated 95th and 99th percentile limits of absolute delta were 12 and 16 ng/L, respectively. We used a conservative cutoff of 16 ng/L as the limit to define a clinically significant increase. Figure 2 View largeDownload slide Absolute (A) and relative (B) differences in high-sensitivity troponin T (hs-TnT) as a function of mean hs-TnT concentrations in the reference cohort. The upper limit of the distribution is more stable across concentrations in A in comparison with the values in B, which shows a progressive drop as concentration increases (n = 810). A, y = 0.125x + 0.25, r = 0.56, P < .001. B, y = 0.135x + 11, r = 0.088, P = .013. Figure 2 View largeDownload slide Absolute (A) and relative (B) differences in high-sensitivity troponin T (hs-TnT) as a function of mean hs-TnT concentrations in the reference cohort. The upper limit of the distribution is more stable across concentrations in A in comparison with the values in B, which shows a progressive drop as concentration increases (n = 810). A, y = 0.125x + 0.25, r = 0.56, P < .001. B, y = 0.135x + 11, r = 0.088, P = .013. We evaluated factors associated with the observed variation in the stable group Table 2. By univariate analysis, age (P < .001), male sex (P = .029), troponin concentration (P < .001), and sampling interval (P < .001) were all associated positively with absolute variability. However, only troponin concentration (P < .001) and sampling interval (P < .001) showed a significant association with variability in multivariate analysis. In patients with at least one hs-TnT more than 14 ng/L, the variability was quite stable when measurements where made 3 hours or more apart Figure 3. The group with a sampling interval below 3 hours showed a lower variability (KW test, P = .002), although the number was small (n = 74). Table 2 Regression Analysis of Variables Associated With High-Sensitivity Troponin T Variability in the Reference Groupa Variable  Univariate  Multivariate  Standard Coefficient  P Value  Standard Coefficient  P Value  Sex (male)  0.037  .029  –0.017  .57  Age (years)  0.22  <.001  –0.047  .15  Mean of troponins  0.56  <.001  0.58  <.001  Sampling interval (hours)  0.14  <.001  0.12  <.001  Variable  Univariate  Multivariate  Standard Coefficient  P Value  Standard Coefficient  P Value  Sex (male)  0.037  .029  –0.017  .57  Age (years)  0.22  <.001  –0.047  .15  Mean of troponins  0.56  <.001  0.58  <.001  Sampling interval (hours)  0.14  <.001  0.12  <.001  aMultivariate model included all four variables. View Large Table 2 Regression Analysis of Variables Associated With High-Sensitivity Troponin T Variability in the Reference Groupa Variable  Univariate  Multivariate  Standard Coefficient  P Value  Standard Coefficient  P Value  Sex (male)  0.037  .029  –0.017  .57  Age (years)  0.22  <.001  –0.047  .15  Mean of troponins  0.56  <.001  0.58  <.001  Sampling interval (hours)  0.14  <.001  0.12  <.001  Variable  Univariate  Multivariate  Standard Coefficient  P Value  Standard Coefficient  P Value  Sex (male)  0.037  .029  –0.017  .57  Age (years)  0.22  <.001  –0.047  .15  Mean of troponins  0.56  <.001  0.58  <.001  Sampling interval (hours)  0.14  <.001  0.12  <.001  aMultivariate model included all four variables. View Large Figure 3 View largeDownload slide Effect of sampling time interval on high-sensitivity troponin T (hs-TnT) differences in the reference cohort by boxplots. Each box represents the 25th and 75th percentiles with the notches showing the confidence interval of the median. A significant difference between medians can be assumed when notches do not overlap, such as between the first two columns. Outliers are represented as asterisks and extreme outliers as circles. Standard criteria to define these outliers are, respectively, any value extending more than 1.5 and 3 times the interquartile range from the median. Figure 3 View largeDownload slide Effect of sampling time interval on high-sensitivity troponin T (hs-TnT) differences in the reference cohort by boxplots. Each box represents the 25th and 75th percentiles with the notches showing the confidence interval of the median. A significant difference between medians can be assumed when notches do not overlap, such as between the first two columns. Outliers are represented as asterisks and extreme outliers as circles. Standard criteria to define these outliers are, respectively, any value extending more than 1.5 and 3 times the interquartile range from the median. In the validation group (n = 2,438) including 396 cases of MI, the area under the ROC curve for the absolute delta was significantly higher than for the first troponin value alone (0.94 vs 0.85, P < .0001; Table 3). An absolute delta of 16 ng/L had a specificity of 94.2% and a sensitivity of 83.2% (Table 3 and Figure 4). Figure 5 illustrates the distribution of patients with a first troponin value below 100 ng/L that would be identified using this criterion (absolute delta >16 ng/L). Table 3 Diagnostic Performances of Different Strategies in the Validation Cohort (n = 2,438) Test  ROC Surface (CI)  Criterion, ng/L  Sensitivity, %  Specificity, %  Troponin 1  0.85a (0.830-0.864)  >14  85.1  59.8      >50  63.4  88.1  Troponin 2  0.95 (0.941-0.958)  >14  97.7  57.5      >50  91.9  86.5  Absolute delta 2-1  0.94 (0.927-0.946)  >16  82.3  91.7  Relative difference %  0.82a (0.805-0.842)  50  58.1  88.1  Test  ROC Surface (CI)  Criterion, ng/L  Sensitivity, %  Specificity, %  Troponin 1  0.85a (0.830-0.864)  >14  85.1  59.8      >50  63.4  88.1  Troponin 2  0.95 (0.941-0.958)  >14  97.7  57.5      >50  91.9  86.5  Absolute delta 2-1  0.94 (0.927-0.946)  >16  82.3  91.7  Relative difference %  0.82a (0.805-0.842)  50  58.1  88.1  CI, confidence interval; ROC, receiver operating characteristic. aSurface significantly different from T2 and absolute difference (P < .0001). Other comparisons: not significant. View Large Table 3 Diagnostic Performances of Different Strategies in the Validation Cohort (n = 2,438) Test  ROC Surface (CI)  Criterion, ng/L  Sensitivity, %  Specificity, %  Troponin 1  0.85a (0.830-0.864)  >14  85.1  59.8      >50  63.4  88.1  Troponin 2  0.95 (0.941-0.958)  >14  97.7  57.5      >50  91.9  86.5  Absolute delta 2-1  0.94 (0.927-0.946)  >16  82.3  91.7  Relative difference %  0.82a (0.805-0.842)  50  58.1  88.1  Test  ROC Surface (CI)  Criterion, ng/L  Sensitivity, %  Specificity, %  Troponin 1  0.85a (0.830-0.864)  >14  85.1  59.8      >50  63.4  88.1  Troponin 2  0.95 (0.941-0.958)  >14  97.7  57.5      >50  91.9  86.5  Absolute delta 2-1  0.94 (0.927-0.946)  >16  82.3  91.7  Relative difference %  0.82a (0.805-0.842)  50  58.1  88.1  CI, confidence interval; ROC, receiver operating characteristic. aSurface significantly different from T2 and absolute difference (P < .0001). Other comparisons: not significant. View Large Figure 4 View largeDownload slide Representation of high-sensitivity troponin T (hs-TnT) 1 and 2 with reference range lines in the validation cohort. The scale for troponin 1 is linear and limited to the range below 100 ng/L where most of the diagnostic challenges are. The y-axis is in logarithmic format to show all troponin 2 results. Straight lines indicate the 99th upper limit of 14 ng/L, while the curved lines represent the difference of ±16 ng/L between troponin 1 and 2. T1 = first hs-TnT, T2 = second hs-TnT, Δ is the absolute difference in ng/L between T1 and T2, and DIFF % is the absolute difference expressed as a percentage of T1. Arrows indicate the coordinate position corresponding to the cutoff of 14 ng/L for T1 and T2, 16 ng/L for Δ, and 30% for DIFF %. ROC, receiver operating characteristic. Figure 4 View largeDownload slide Representation of high-sensitivity troponin T (hs-TnT) 1 and 2 with reference range lines in the validation cohort. The scale for troponin 1 is linear and limited to the range below 100 ng/L where most of the diagnostic challenges are. The y-axis is in logarithmic format to show all troponin 2 results. Straight lines indicate the 99th upper limit of 14 ng/L, while the curved lines represent the difference of ±16 ng/L between troponin 1 and 2. T1 = first hs-TnT, T2 = second hs-TnT, Δ is the absolute difference in ng/L between T1 and T2, and DIFF % is the absolute difference expressed as a percentage of T1. Arrows indicate the coordinate position corresponding to the cutoff of 14 ng/L for T1 and T2, 16 ng/L for Δ, and 30% for DIFF %. ROC, receiver operating characteristic. Figure 5 View largeDownload slide Representation of the first high-sensitivity troponin T (hs-TnT) and the second hs-TnT with reference range lines in the validation cohort (n = 2,348). The scale for hs-TnT 1 is linear and limited to the range below 100 ng/L where most of the diagnostic challenges are. The y-axis is in logarithmic format to show all second hs-TnT results. Straight lines indicate the 99th upper limit of 14 ng/L, while the curved lines represent the delta of ±16 ng/L between hs-TnT 1 and 2. Data points outside these lines are considered a rule-in criterion for acute heart injury (myocardial infarction diagnosis shown as circles). Figure 5 View largeDownload slide Representation of the first high-sensitivity troponin T (hs-TnT) and the second hs-TnT with reference range lines in the validation cohort (n = 2,348). The scale for hs-TnT 1 is linear and limited to the range below 100 ng/L where most of the diagnostic challenges are. The y-axis is in logarithmic format to show all second hs-TnT results. Straight lines indicate the 99th upper limit of 14 ng/L, while the curved lines represent the delta of ±16 ng/L between hs-TnT 1 and 2. Data points outside these lines are considered a rule-in criterion for acute heart injury (myocardial infarction diagnosis shown as circles). Discussion In a large retrospective study in patients presenting to the emergency room, we derived a threshold for the absolute delta in two serial measurements of hs-TnT to identify patients with meaningful change suggestive of MI or other acute events. To our knowledge, our cohort is the largest to consider biological variation in the critical zone. For patients with values between 14 and 50 ng/L, a delta of 16 ng/L was highly predictive of MI and was not influenced by age and sex. Globally, we observed a high prevalence (32%) of abnormal hs-TnT for the first measurement, consistent with previous reports. In ambulatory elderly patients without acute conditions (age older than 75 years), the prevalence of hs-TnT above 14 ng/L has been reported between 40% and 50%.9-11 Thus, the cutoff of 14 ng/L provides limited specificity. Uncertainty in the low abnormal range can best be resolved by considering sequential samples. Our results fully support that absolute deltas in troponin as seen in Figure 4 should be preferred for diagnostic purposes.17-20 Strangely, some authors have focused solely on the analytical variation to recommend low relative cutoffs, such as 20%, to identify a significant increase. This would require no variation due to sampling and no oscillation in plasma. Other authors have suggested limits between 20% and 112% from limited numbers of patients.12-15 Our data clearly show that it is not possible to define a single cutoff for a relative change in percentage as the limit decreases progressively with increasing concentrations of hs-TnT (Figure 2). We suggest that changes in hs-TnT should only be expressed as percentages in the range above 80 ng/L. Since the diagnosis of MI requires at least one hs-TnT over 14 ng/L, we were especially interested in the stable patients with one value over 14 ng/L. Our observed limit is larger than those found in studies with healthy volunteers. For example, one study reported an upper 95th percentile limit of 8.3 ng/L for the maximum delta in observed troponins in 67 patients.20 However, 53 of 67 patients had values of 14 ng/L or less. Thus, this limit underestimates the variation for patients with troponin in the critical zone over 14 ng/L. In our opinion, studying variability in healthy individuals with very low troponin levels has little relevance to older patients with chronically slightly abnormal levels. Although the absolute variation increases with average concentrations, an upper uniform limit of 16 ng/L is a reasonable option for concentrations between 15 and 80 ng/L. A delta of hs-TnT exceeding 16 ng/L (increasing or decreasing) is therefore an unusual occurrence in stable patients and is compatible with an acute event (MI or other diagnosis). It is interesting to note that sex and age are not significant contributors of normal short-term troponin variability in multivariate analysis. This suggests that when two results are available, interpretation of the delta should not be influenced by sex or age. This is a welcome conclusion since specific criteria are not required. In other words, the same limits of absolute delta can be used regardless of sex and age for the interpretation of sequential results. The threshold for absolute delta was stable across different sampling intervals between 3 and 24 hours, suggesting that the periodicity of biological oscillation in stable patients is below a few hours. Similar observations were reported for high-sensitivity troponin I in a recent study of biological variation in patients presenting to the emergency department.22 Data for sampling intervals below 3 hours were limited, but it is very likely that smaller cutoffs should be used for optimal strategic management. Recent studies have proposed different algorithms for ruling out and ruling in MI using sampling intervals as low as 1 hour.16,23 Recommendations to use a delta below 7 ng/L as a rule-out criterion to identify a low-risk group is a reasonable strategy.17,18,21,22 Here, we suggest that a delta over 16 ng/L in the low range is a proper rule-in criterion with a high rate of acute events. Patients falling between 7 and 16 are in a gray zone of uncertainly and should be further investigated with additional testing. A limitation of our study is that some patients with undiagnosed MI and decreasing hs-TnT values could have been included in the reference group and introduced some bias in the observed cutoffs. Furthermore, we relied on the acting physicians for the final diagnosis of MI, which was most likely influenced by the hs-TnT results, although nearly all patients with MI subsequently underwent imaging modalities as the hospital is a cardiac referral center. We could not assess precisely the timing of the MI with respect to the first two samples, and this might have reduced the observed sensitivity and specificity. We could not assess the diagnostic performance for non-ST elevation MI and ST elevation MI specifically as our validation data set does not include that information. Furthermore, another limitation is that we did not try to evaluate delta as a prognostic factor for mortality or other major coronary events at 1 month or 1 year. Globally, our results are consistent with other reports and allow comparisons between potential strategies. Since we used a large pool of real patients’ data, we report robust evaluations that take into account the biological and analytical variations in a real clinical setting. Conclusion The absolute delta in hs-TnT is a dynamic reflection of the evolution of a potential acute event. Based on the variation observed in a clinical setting, we propose that an absolute cutoff of 16 ng/L should be used as a specific limit (rule-in criterion) for the delta in serial measurements with a sampling interval of 3 hours or more to identify acute myocardial events. This criterion is proposed as a practical and easy way to interpret results in an emergency setting to identify significant alterations in hs-TnT in patients with moderately elevated values, independently of age and sex. Further studies are needed for fast protocols that include resampling before 3 hours since biological variability is expected to be lower within such a short time frame. Pierre Douville receives a grant from Roche Diagnostics for a project unrelated to this report. References 1. de Lemos JA, Drazner MH, Omland T, et al.   Association of troponin T detected with a highly sensitive assay and cardiac structure and mortality risk in the general population. JAMA . 2010; 304: 2503- 2512. Google Scholar CrossRef Search ADS PubMed  2. Gore MO, Seliger SL, Defilippi CR, et al.   Age- and sex-dependent upper reference limits for the high-sensitivity cardiac troponin T assay. J Am Coll Cardiol . 2014; 63: 1441- 1448. Google Scholar CrossRef Search ADS PubMed  3. Eggers KM, Lind L, Venge P, et al.   Factors influencing the 99th percentile of cardiac troponin I evaluated in community-dwelling individuals at 70 and 75 years of age. Clin Chem . 2013; 59: 1068- 1073. Google Scholar CrossRef Search ADS PubMed  4. Biener M, Giannitsis E, Lamerz J, et al.   Prognostic value of elevated high-sensitivity cardiac troponin T levels in a low risk outpatient population with cardiovascular disease. Eur Heart J Acute Cardiovasc Care . 2016; 5: 409- 418. Google Scholar CrossRef Search ADS PubMed  5. Franzini M, Lorenzoni V, Masotti S, et al.   The calculation of the cardiac troponin T 99th percentile of the reference population is affected by age, gender, and population selection: a multicenter study in Italy. Clin Chim Acta . 2015; 438: 376- 381. Google Scholar CrossRef Search ADS PubMed  6. Chenevier-Gobeaux C, Bailleul S, Mzabi A, et al.   Upper reference limits of high-sensitivity cardiac troponin T in a general population: comparison with those of sensitive cardiac troponin I. Clin Lab . 2013; 59: 333- 336. Google Scholar CrossRef Search ADS PubMed  7. Roffi M, Patrono C, Collet JP, et al.  ; Management of Acute Coronary Syndromes in Patients Presenting Without Persistent ST-Segment Elevation of the European Society of Cardiology. 2015 ESC guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation: task force for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation of the European Society of Cardiology (ESC). Eur Heart J . 2016; 37: 267- 315. Google Scholar CrossRef Search ADS PubMed  8. McMullin N, Lindsell CJ, Lei L, et al.  ; EMCREG i*trACS Investigators. Outcomes associated with small changes in normal-range cardiac markers. Am J Emerg Med . 2011; 29: 162- 167. Google Scholar CrossRef Search ADS PubMed  9. Menacer S, Claessens YE, Meune C, et al.   Reference range values of troponin measured by sensitive assays in elderly patients without any cardiac signs/symptoms. Clin Chim Acta . 2013; 417: 45- 47. Google Scholar CrossRef Search ADS PubMed  10. Mansour M, Clark L, Scott T, et al.   Considerations for establishing a reference interval for elderly individuals in the emergency department with the high-sensitivity cardiac troponin T assay. Clin Chim Acta . 2013; 421: 85- 86. Google Scholar CrossRef Search ADS PubMed  11. Sajeev JK, New G, Roberts L, et al.   High sensitivity troponin: does the 50% delta change alter clinical outcomes in chest pain presentations to the emergency room? Int J Cardiol . 2015; 184: 170- 174. Google Scholar CrossRef Search ADS PubMed  12. Buccelletti F, Galiuto L, Marsiliani D, et al.   Comparison of diagnostic accuracy between three different rules of interpreting high sensitivity troponin T results. Intern Emerg Med . 2012; 7: 365- 370. Google Scholar CrossRef Search ADS PubMed  13. Frankenstein L, Wu AH, Hallermayer K, et al.   Biological variation and reference change value of high-sensitivity troponin T in healthy individuals during short and intermediate follow-up periods. Clin Chem . 2011; 57: 1068- 1071. Google Scholar CrossRef Search ADS PubMed  14. Vasile VC, Saenger AK, Kroning JM, et al.   Biological and analytical variability of a novel high-sensitivity cardiac troponin T assay. Clin Chem . 2010; 56: 1086- 1090. Google Scholar CrossRef Search ADS PubMed  15. Nordenskjöld AM, Ahlström H, Eggers KM, et al.   Short- and long-term individual variation in cardiac troponin in patients with stable coronary artery disease. Clin Chem . 2013; 59: 401- 409. Google Scholar CrossRef Search ADS PubMed  16. Reichlin T, Schindler C, Drexler B, et al.   One-hour rule-out and rule-in of acute myocardial infarction using high-sensitivity cardiac troponin T. Arch Intern Med . 2012; 172: 1211- 1218. Google Scholar CrossRef Search ADS PubMed  17. Biener M, Mueller M, Vafaie M, et al.   Comparison of a 3-hour versus a 6-hour sampling-protocol using high-sensitivity cardiac troponin T for rule-out and rule-in of non-STEMI in an unselected emergency department population. Int J Cardiol . 2013; 167: 1134- 1140. Google Scholar CrossRef Search ADS PubMed  18. Mueller M, Biener M, Vafaie M, et al.   Absolute and relative kinetic changes of high-sensitivity cardiac troponin T in acute coronary syndrome and in patients with increased troponin in the absence of acute coronary syndrome. Clin Chem . 2012; 58: 209- 218. Google Scholar CrossRef Search ADS PubMed  19. Reichlin T, Irfan A, Twerenbold R, et al.   Utility of absolute and relative changes in cardiac troponin concentrations in the early diagnosis of acute myocardial infarction. Circulation . 2011; 124: 136- 145. Google Scholar CrossRef Search ADS PubMed  20. Pretorius CJ, Wilgen U, Ungerer JP. Serial cardiac troponin differences measured on four contemporary analyzers: relative differences, actual differences and reference change values compared. Clin Chim Acta . 2012; 413: 1786- 1791. Google Scholar CrossRef Search ADS PubMed  21. Scharnhorst V, Krasznai K, van’t Veer M, et al.   Variation of cardiac troponin I and T measured with sensitive assays in emergency department patients with noncardiac chest pain. Clin Chem . 2012; 58: 1208- 1214. Google Scholar CrossRef Search ADS PubMed  22. Simpson AJ, Potter JM, Koerbin G, et al.   Use of observed within-person variation of cardiac troponin in emergency department patients for determination of biological variation and percentage and absolute reference change values. Clin Chem . 2014; 60: 848- 854. Google Scholar CrossRef Search ADS PubMed  23. Reichlin T, Cullen L, Parsonage WA, et al.   Two-hour algorithm for triage toward rule-out and rule-in of acute myocardial infarction using high-sensitivity cardiac troponin T. Am J Med . 2015; 128: 369- 379.e4. Google Scholar CrossRef Search ADS PubMed  © American Society for Clinical Pathology, 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png American Journal of Clinical Pathology Oxford University Press

Variability of High-Sensitivity Troponin T Concentrations in Emergency Settings: Impact for the Diagnosis of Myocardial Infarction

Loading next page...
 
/lp/ou_press/variability-of-high-sensitivity-troponin-t-concentrations-in-emergency-vZ62ibCkah
Publisher
Oxford University Press
Copyright
© American Society for Clinical Pathology, 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0002-9173
eISSN
1943-7722
D.O.I.
10.1093/ajcp/aqy025
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives To assess biological variation of troponin T in emergency settings and establish limits for interpretation of serial results. Methods We studied 6,557 consecutive patients with troponin measurements. A stable reference subset was selected to estimate biological variation and threshold limits. Results The first troponin level was elevated in 32% of patients, and 2,490 had a second troponin level with a myocardial infarction (MI) prevalence of 16.2%. In the stable reference group with at least one abnormal value, the 99th percentile of the absolute delta between the first two samples was 16 ng/L. For MI diagnosis, the area under the receiver operating characteristic curve was 0.85 (confidence interval [CI], 0.83-0.87) for the first troponin level and 0.94 (CI, 0.93-0.95) for the absolute delta. Conclusions An absolute delta of 16 ng/L has good specificity in the emergency setting. This threshold is valid for any sex, age, and sampling interval between 3 and 24 hours and is higher than published limits found in healthy outpatients. Myocardial infarction, Troponin, Acute coronary syndrome, Biological variation, Reference range Many individuals without acute conditions have basal high-sensitivity troponin T (hs-TnT) values repeatedly above 14 ng/L, the recommended 99th percentile threshold for the diagnosis of myocardial infarction (MI). Factors associated with increased values include older age, male sex, chronic heart disease, peripheral atherosclerotic vascular disease, hypertension, diabetes, and chronic kidney disease.1-4 For patients with chronic elevations of troponin T, the challenge is to differentiate “stable” patients from patients with an acute condition that increases troponin, especially an MI. Some authors have suggested to use age and sex-adjusted upper limits to improve discrimination.2,5,6 However, dynamic changes are still required in most cases to identify MI with certainty.7 Moreover, patients presenting to the emergency room in a clinical context are more likely to have comorbidities and have a chronic elevation of hs-TnT than typically studied reference populations with young and healthy individuals. Our objectives were to assess biological variability of hs-TnT in unselected patients presenting to the emergency room and define a 99th percentile limit for the delta in sequential troponin results in stable patients without acute injury. This limit could then be applied as a rule-in criterion to identify acute cardiac injury. Secondary objectives were to evaluate the diagnostic accuracy of the criterion to recognize MI and to identify factors associated with increased biological variability. Materials and Methods We performed a retrospective study of consecutive patients presenting to the emergency department of a university hospital (IUCPQ, Québec, Canada). In total, 6,557 consecutive patients with at least one hs-TnT measurement were included. Only the first encounter was studied for each patient. hs-TnT was measured in heparinized plasma on a stat basis with modular analyzers (Roche Diagnostics, Montreal, Canada) according to the manufacturer’s instructions. Imprecision is below 10% for values above 10 ng/L. All results are expressed as whole numbers without decimals. We predefined the following cutoffs: (1) 3 ng/L as the limit of detection, with any value below 3 set at 3 ng/L for statistical analyses; (2) 14 ng/L as the upper limit of normal, as suggested by the manufacturer; and (3) 50 ng/L as equivalent to the upper reference limit used with the previous troponin T assay to identify a high-risk group. Variation of troponins was analyzed in patients with repeated troponin measurements within a limit of 24 hours (n = 2,490). Our strategy was to select a reference cohort from the whole cohort with a very low probability of acute events, apply the principle of symmetry as described below, determine a 99th percentile limit of variation, and evaluate its performance in the whole cohort. We focused our analysis on patients with normal or slightly abnormal (below 50 ng/L) levels since this is the critical zone where sequential analysis is most needed. Patients with MI or other acute events typically show increases in troponin. In contrast, patients without acute events may show either increasing or decreasing sequential values in a random manner. Patients who have no change (identical results) in troponin or decreasing values with an average below 50 ng/L are at very low risk of MI. McMullin et al8 found that a decreasing troponin T in an emergency setting was not associated with increased odds of acute coronary syndrome in a cohort of over 1,800 patients. Among patients with a mean hs-TnT less than 50 ng/L (n = 1,861), we thus defined a reference cohort of stable patients with equal or decreasing results. For stable patients, it is expected that hs-TnT values vary randomly around a mean such that for two sequential results, there are equal chances to have R1 > R2 or R2 > R1, with half going up and half going down. Thus, the absolute variation can be used as a surrogate of variability in stable low-risk patients (principle of symmetry). To avoid statistical bias, we included only half of the patients with identical values selected randomly and excluded extreme outliers defined as a change exceeding three times the interquartile range (a standard criterion for extreme outliers). Figure 1 describes the selection process for the reference cohort. Figure 1 View largeDownload slide Cohort selection flowchart. Figure 1 View largeDownload slide Cohort selection flowchart. We evaluated the diagnostic performance of the threshold found in the reference cohort study for the identification of MI. In the absence of an independent gold standard, we used the final diagnosis from the electronic medical records as established by the treating physicians without any patient chart reviews. Complementary investigations were at the discretion of the clinicians, and all troponin results were available to physicians for their final judgment. The hospital research ethics review board approved the study. The validation cohort included patients with at least two hs-TnT measurements within a 24-hour period (Figure 1; n = 2,490). Because of the uncertainty about troponin evolution, patients with a diagnosis of cardiomyositis, pericardial disease, or circulatory shock were excluded (n = 52). All others (n = 2,438) were included in the validation cohort and were divided into positive or negative for MI. Patients with cardiac arrest were included in the MI group (Figure 1). Statistical Analyses We made comparisons between groups with the standard t test, the χ2 test, and the nonparametric Kruskal-Wallis (KW) test or Mann-Whitney U test where appropriate. Outliers and extreme outliers were defined as, respectively, any value exceeding 1.5 and 3 times the interquartile range from the median. In the reference group, we defined the threshold to identify a significant change in hs-TnT based on the 99th percentile of the delta distribution between repeated measures. Variables associated with the absolute delta between two consecutive troponin results were analyzed by univariate and multivariate linear regression statistics in the reference group. To evaluate the performance of each testing strategy, we calculated sensitivity, specificity, and area under the receiver operating characteristic (ROC) curve. Logarithmic scale was used to better visualize results in the low range. Results We identified 6,557 consecutive patients with an hs-TnT measurement in the emergency department. Median age was 68 years, and 55% were men. There were 435 MI cases, representing 6.6% of patients. The first hs-TnT was normal (≤14 ng/L) in 68%, slightly elevated (between 15 and 50 ng/L) in 23%, and above 50 ng/L in 9%. In those three groups, MI was diagnosed respectively in 1.4%, 6.5%, and 46.6%. Of the whole cohort, 38% had a second sample within 24 hours (n = 2,490 with 16.2% MI). Of those, 1,861 had an average troponin below 50 ng/L, among which 445 had decreasing and 751 had identical values. Figure 1 and Table 1 describe the population and the subgroups studied. To select the reference stable group, we kept the patients with decreasing values (n = 445) and added half of the 751 patients with equal values selected randomly (n = 376), thus generating a set of 810 patients after exclusion of 11 extreme outliers (delta above 20 ng/L). Basic characteristics of these groups are displayed in Table 1. Differences between group B and the reference group are statistically significant for age, sex, and troponin distributions (P < .01), mainly due to the large sample size. However, we consider the reference group quite representative of the population served (Table 1). Table 1 Cohort Characteristics Groupa  Complete Cohort  Validation Cohort  Reference Cohort  No. of patients  6,557  2,438  810  Age, median, ya  68.3  70.4  68.8  Age, mean ± SD, ya  66 ± 16  69.4 ± 14  67.8 ± 14  Sex (male), %  55.4  58.9  55.2  First troponin T, median, ng/La  6  13  7   0-14 ng/L, %a  68  52.5  65.8   15-50 ng/L, %a  22.6  27.2  32.6   >50 ng/L, %a  9.4  20.3  1.6  Second troponin T, median, ng/La  —  16  5  Sampling interval, median, ha  —  6.7  6.2  Groupa  Complete Cohort  Validation Cohort  Reference Cohort  No. of patients  6,557  2,438  810  Age, median, ya  68.3  70.4  68.8  Age, mean ± SD, ya  66 ± 16  69.4 ± 14  67.8 ± 14  Sex (male), %  55.4  58.9  55.2  First troponin T, median, ng/La  6  13  7   0-14 ng/L, %a  68  52.5  65.8   15-50 ng/L, %a  22.6  27.2  32.6   >50 ng/L, %a  9.4  20.3  1.6  Second troponin T, median, ng/La  —  16  5  Sampling interval, median, ha  —  6.7  6.2  —, data not available because of absence of second sample in many cases. aComparison between cohorts is statistically significant (P < .01). View Large Table 1 Cohort Characteristics Groupa  Complete Cohort  Validation Cohort  Reference Cohort  No. of patients  6,557  2,438  810  Age, median, ya  68.3  70.4  68.8  Age, mean ± SD, ya  66 ± 16  69.4 ± 14  67.8 ± 14  Sex (male), %  55.4  58.9  55.2  First troponin T, median, ng/La  6  13  7   0-14 ng/L, %a  68  52.5  65.8   15-50 ng/L, %a  22.6  27.2  32.6   >50 ng/L, %a  9.4  20.3  1.6  Second troponin T, median, ng/La  —  16  5  Sampling interval, median, ha  —  6.7  6.2  Groupa  Complete Cohort  Validation Cohort  Reference Cohort  No. of patients  6,557  2,438  810  Age, median, ya  68.3  70.4  68.8  Age, mean ± SD, ya  66 ± 16  69.4 ± 14  67.8 ± 14  Sex (male), %  55.4  58.9  55.2  First troponin T, median, ng/La  6  13  7   0-14 ng/L, %a  68  52.5  65.8   15-50 ng/L, %a  22.6  27.2  32.6   >50 ng/L, %a  9.4  20.3  1.6  Second troponin T, median, ng/La  —  16  5  Sampling interval, median, ha  —  6.7  6.2  —, data not available because of absence of second sample in many cases. aComparison between cohorts is statistically significant (P < .01). View Large The absolute delta between the two measurements increased progressively (r = 0.56, P < .001) with increasing hs-TnT Figure 2A. However, for patients with values between 15 and 50 ng/L, the upper limit of dispersion was relatively stable, suggesting that a unique cutoff for absolute delta could be used. Relative change showed a progressive decrease in dispersion with increasing hs-TnT values Figure 2B. For the reference group of 810 patients, the 99th percentile of the absolute delta between samples 1 and 2 was 14 ng/L. However, for subjects with at least one value above 14 ng/L (a condition required for the diagnosis of MI), the calculated 95th and 99th percentile limits of absolute delta were 12 and 16 ng/L, respectively. We used a conservative cutoff of 16 ng/L as the limit to define a clinically significant increase. Figure 2 View largeDownload slide Absolute (A) and relative (B) differences in high-sensitivity troponin T (hs-TnT) as a function of mean hs-TnT concentrations in the reference cohort. The upper limit of the distribution is more stable across concentrations in A in comparison with the values in B, which shows a progressive drop as concentration increases (n = 810). A, y = 0.125x + 0.25, r = 0.56, P < .001. B, y = 0.135x + 11, r = 0.088, P = .013. Figure 2 View largeDownload slide Absolute (A) and relative (B) differences in high-sensitivity troponin T (hs-TnT) as a function of mean hs-TnT concentrations in the reference cohort. The upper limit of the distribution is more stable across concentrations in A in comparison with the values in B, which shows a progressive drop as concentration increases (n = 810). A, y = 0.125x + 0.25, r = 0.56, P < .001. B, y = 0.135x + 11, r = 0.088, P = .013. We evaluated factors associated with the observed variation in the stable group Table 2. By univariate analysis, age (P < .001), male sex (P = .029), troponin concentration (P < .001), and sampling interval (P < .001) were all associated positively with absolute variability. However, only troponin concentration (P < .001) and sampling interval (P < .001) showed a significant association with variability in multivariate analysis. In patients with at least one hs-TnT more than 14 ng/L, the variability was quite stable when measurements where made 3 hours or more apart Figure 3. The group with a sampling interval below 3 hours showed a lower variability (KW test, P = .002), although the number was small (n = 74). Table 2 Regression Analysis of Variables Associated With High-Sensitivity Troponin T Variability in the Reference Groupa Variable  Univariate  Multivariate  Standard Coefficient  P Value  Standard Coefficient  P Value  Sex (male)  0.037  .029  –0.017  .57  Age (years)  0.22  <.001  –0.047  .15  Mean of troponins  0.56  <.001  0.58  <.001  Sampling interval (hours)  0.14  <.001  0.12  <.001  Variable  Univariate  Multivariate  Standard Coefficient  P Value  Standard Coefficient  P Value  Sex (male)  0.037  .029  –0.017  .57  Age (years)  0.22  <.001  –0.047  .15  Mean of troponins  0.56  <.001  0.58  <.001  Sampling interval (hours)  0.14  <.001  0.12  <.001  aMultivariate model included all four variables. View Large Table 2 Regression Analysis of Variables Associated With High-Sensitivity Troponin T Variability in the Reference Groupa Variable  Univariate  Multivariate  Standard Coefficient  P Value  Standard Coefficient  P Value  Sex (male)  0.037  .029  –0.017  .57  Age (years)  0.22  <.001  –0.047  .15  Mean of troponins  0.56  <.001  0.58  <.001  Sampling interval (hours)  0.14  <.001  0.12  <.001  Variable  Univariate  Multivariate  Standard Coefficient  P Value  Standard Coefficient  P Value  Sex (male)  0.037  .029  –0.017  .57  Age (years)  0.22  <.001  –0.047  .15  Mean of troponins  0.56  <.001  0.58  <.001  Sampling interval (hours)  0.14  <.001  0.12  <.001  aMultivariate model included all four variables. View Large Figure 3 View largeDownload slide Effect of sampling time interval on high-sensitivity troponin T (hs-TnT) differences in the reference cohort by boxplots. Each box represents the 25th and 75th percentiles with the notches showing the confidence interval of the median. A significant difference between medians can be assumed when notches do not overlap, such as between the first two columns. Outliers are represented as asterisks and extreme outliers as circles. Standard criteria to define these outliers are, respectively, any value extending more than 1.5 and 3 times the interquartile range from the median. Figure 3 View largeDownload slide Effect of sampling time interval on high-sensitivity troponin T (hs-TnT) differences in the reference cohort by boxplots. Each box represents the 25th and 75th percentiles with the notches showing the confidence interval of the median. A significant difference between medians can be assumed when notches do not overlap, such as between the first two columns. Outliers are represented as asterisks and extreme outliers as circles. Standard criteria to define these outliers are, respectively, any value extending more than 1.5 and 3 times the interquartile range from the median. In the validation group (n = 2,438) including 396 cases of MI, the area under the ROC curve for the absolute delta was significantly higher than for the first troponin value alone (0.94 vs 0.85, P < .0001; Table 3). An absolute delta of 16 ng/L had a specificity of 94.2% and a sensitivity of 83.2% (Table 3 and Figure 4). Figure 5 illustrates the distribution of patients with a first troponin value below 100 ng/L that would be identified using this criterion (absolute delta >16 ng/L). Table 3 Diagnostic Performances of Different Strategies in the Validation Cohort (n = 2,438) Test  ROC Surface (CI)  Criterion, ng/L  Sensitivity, %  Specificity, %  Troponin 1  0.85a (0.830-0.864)  >14  85.1  59.8      >50  63.4  88.1  Troponin 2  0.95 (0.941-0.958)  >14  97.7  57.5      >50  91.9  86.5  Absolute delta 2-1  0.94 (0.927-0.946)  >16  82.3  91.7  Relative difference %  0.82a (0.805-0.842)  50  58.1  88.1  Test  ROC Surface (CI)  Criterion, ng/L  Sensitivity, %  Specificity, %  Troponin 1  0.85a (0.830-0.864)  >14  85.1  59.8      >50  63.4  88.1  Troponin 2  0.95 (0.941-0.958)  >14  97.7  57.5      >50  91.9  86.5  Absolute delta 2-1  0.94 (0.927-0.946)  >16  82.3  91.7  Relative difference %  0.82a (0.805-0.842)  50  58.1  88.1  CI, confidence interval; ROC, receiver operating characteristic. aSurface significantly different from T2 and absolute difference (P < .0001). Other comparisons: not significant. View Large Table 3 Diagnostic Performances of Different Strategies in the Validation Cohort (n = 2,438) Test  ROC Surface (CI)  Criterion, ng/L  Sensitivity, %  Specificity, %  Troponin 1  0.85a (0.830-0.864)  >14  85.1  59.8      >50  63.4  88.1  Troponin 2  0.95 (0.941-0.958)  >14  97.7  57.5      >50  91.9  86.5  Absolute delta 2-1  0.94 (0.927-0.946)  >16  82.3  91.7  Relative difference %  0.82a (0.805-0.842)  50  58.1  88.1  Test  ROC Surface (CI)  Criterion, ng/L  Sensitivity, %  Specificity, %  Troponin 1  0.85a (0.830-0.864)  >14  85.1  59.8      >50  63.4  88.1  Troponin 2  0.95 (0.941-0.958)  >14  97.7  57.5      >50  91.9  86.5  Absolute delta 2-1  0.94 (0.927-0.946)  >16  82.3  91.7  Relative difference %  0.82a (0.805-0.842)  50  58.1  88.1  CI, confidence interval; ROC, receiver operating characteristic. aSurface significantly different from T2 and absolute difference (P < .0001). Other comparisons: not significant. View Large Figure 4 View largeDownload slide Representation of high-sensitivity troponin T (hs-TnT) 1 and 2 with reference range lines in the validation cohort. The scale for troponin 1 is linear and limited to the range below 100 ng/L where most of the diagnostic challenges are. The y-axis is in logarithmic format to show all troponin 2 results. Straight lines indicate the 99th upper limit of 14 ng/L, while the curved lines represent the difference of ±16 ng/L between troponin 1 and 2. T1 = first hs-TnT, T2 = second hs-TnT, Δ is the absolute difference in ng/L between T1 and T2, and DIFF % is the absolute difference expressed as a percentage of T1. Arrows indicate the coordinate position corresponding to the cutoff of 14 ng/L for T1 and T2, 16 ng/L for Δ, and 30% for DIFF %. ROC, receiver operating characteristic. Figure 4 View largeDownload slide Representation of high-sensitivity troponin T (hs-TnT) 1 and 2 with reference range lines in the validation cohort. The scale for troponin 1 is linear and limited to the range below 100 ng/L where most of the diagnostic challenges are. The y-axis is in logarithmic format to show all troponin 2 results. Straight lines indicate the 99th upper limit of 14 ng/L, while the curved lines represent the difference of ±16 ng/L between troponin 1 and 2. T1 = first hs-TnT, T2 = second hs-TnT, Δ is the absolute difference in ng/L between T1 and T2, and DIFF % is the absolute difference expressed as a percentage of T1. Arrows indicate the coordinate position corresponding to the cutoff of 14 ng/L for T1 and T2, 16 ng/L for Δ, and 30% for DIFF %. ROC, receiver operating characteristic. Figure 5 View largeDownload slide Representation of the first high-sensitivity troponin T (hs-TnT) and the second hs-TnT with reference range lines in the validation cohort (n = 2,348). The scale for hs-TnT 1 is linear and limited to the range below 100 ng/L where most of the diagnostic challenges are. The y-axis is in logarithmic format to show all second hs-TnT results. Straight lines indicate the 99th upper limit of 14 ng/L, while the curved lines represent the delta of ±16 ng/L between hs-TnT 1 and 2. Data points outside these lines are considered a rule-in criterion for acute heart injury (myocardial infarction diagnosis shown as circles). Figure 5 View largeDownload slide Representation of the first high-sensitivity troponin T (hs-TnT) and the second hs-TnT with reference range lines in the validation cohort (n = 2,348). The scale for hs-TnT 1 is linear and limited to the range below 100 ng/L where most of the diagnostic challenges are. The y-axis is in logarithmic format to show all second hs-TnT results. Straight lines indicate the 99th upper limit of 14 ng/L, while the curved lines represent the delta of ±16 ng/L between hs-TnT 1 and 2. Data points outside these lines are considered a rule-in criterion for acute heart injury (myocardial infarction diagnosis shown as circles). Discussion In a large retrospective study in patients presenting to the emergency room, we derived a threshold for the absolute delta in two serial measurements of hs-TnT to identify patients with meaningful change suggestive of MI or other acute events. To our knowledge, our cohort is the largest to consider biological variation in the critical zone. For patients with values between 14 and 50 ng/L, a delta of 16 ng/L was highly predictive of MI and was not influenced by age and sex. Globally, we observed a high prevalence (32%) of abnormal hs-TnT for the first measurement, consistent with previous reports. In ambulatory elderly patients without acute conditions (age older than 75 years), the prevalence of hs-TnT above 14 ng/L has been reported between 40% and 50%.9-11 Thus, the cutoff of 14 ng/L provides limited specificity. Uncertainty in the low abnormal range can best be resolved by considering sequential samples. Our results fully support that absolute deltas in troponin as seen in Figure 4 should be preferred for diagnostic purposes.17-20 Strangely, some authors have focused solely on the analytical variation to recommend low relative cutoffs, such as 20%, to identify a significant increase. This would require no variation due to sampling and no oscillation in plasma. Other authors have suggested limits between 20% and 112% from limited numbers of patients.12-15 Our data clearly show that it is not possible to define a single cutoff for a relative change in percentage as the limit decreases progressively with increasing concentrations of hs-TnT (Figure 2). We suggest that changes in hs-TnT should only be expressed as percentages in the range above 80 ng/L. Since the diagnosis of MI requires at least one hs-TnT over 14 ng/L, we were especially interested in the stable patients with one value over 14 ng/L. Our observed limit is larger than those found in studies with healthy volunteers. For example, one study reported an upper 95th percentile limit of 8.3 ng/L for the maximum delta in observed troponins in 67 patients.20 However, 53 of 67 patients had values of 14 ng/L or less. Thus, this limit underestimates the variation for patients with troponin in the critical zone over 14 ng/L. In our opinion, studying variability in healthy individuals with very low troponin levels has little relevance to older patients with chronically slightly abnormal levels. Although the absolute variation increases with average concentrations, an upper uniform limit of 16 ng/L is a reasonable option for concentrations between 15 and 80 ng/L. A delta of hs-TnT exceeding 16 ng/L (increasing or decreasing) is therefore an unusual occurrence in stable patients and is compatible with an acute event (MI or other diagnosis). It is interesting to note that sex and age are not significant contributors of normal short-term troponin variability in multivariate analysis. This suggests that when two results are available, interpretation of the delta should not be influenced by sex or age. This is a welcome conclusion since specific criteria are not required. In other words, the same limits of absolute delta can be used regardless of sex and age for the interpretation of sequential results. The threshold for absolute delta was stable across different sampling intervals between 3 and 24 hours, suggesting that the periodicity of biological oscillation in stable patients is below a few hours. Similar observations were reported for high-sensitivity troponin I in a recent study of biological variation in patients presenting to the emergency department.22 Data for sampling intervals below 3 hours were limited, but it is very likely that smaller cutoffs should be used for optimal strategic management. Recent studies have proposed different algorithms for ruling out and ruling in MI using sampling intervals as low as 1 hour.16,23 Recommendations to use a delta below 7 ng/L as a rule-out criterion to identify a low-risk group is a reasonable strategy.17,18,21,22 Here, we suggest that a delta over 16 ng/L in the low range is a proper rule-in criterion with a high rate of acute events. Patients falling between 7 and 16 are in a gray zone of uncertainly and should be further investigated with additional testing. A limitation of our study is that some patients with undiagnosed MI and decreasing hs-TnT values could have been included in the reference group and introduced some bias in the observed cutoffs. Furthermore, we relied on the acting physicians for the final diagnosis of MI, which was most likely influenced by the hs-TnT results, although nearly all patients with MI subsequently underwent imaging modalities as the hospital is a cardiac referral center. We could not assess precisely the timing of the MI with respect to the first two samples, and this might have reduced the observed sensitivity and specificity. We could not assess the diagnostic performance for non-ST elevation MI and ST elevation MI specifically as our validation data set does not include that information. Furthermore, another limitation is that we did not try to evaluate delta as a prognostic factor for mortality or other major coronary events at 1 month or 1 year. Globally, our results are consistent with other reports and allow comparisons between potential strategies. Since we used a large pool of real patients’ data, we report robust evaluations that take into account the biological and analytical variations in a real clinical setting. Conclusion The absolute delta in hs-TnT is a dynamic reflection of the evolution of a potential acute event. Based on the variation observed in a clinical setting, we propose that an absolute cutoff of 16 ng/L should be used as a specific limit (rule-in criterion) for the delta in serial measurements with a sampling interval of 3 hours or more to identify acute myocardial events. This criterion is proposed as a practical and easy way to interpret results in an emergency setting to identify significant alterations in hs-TnT in patients with moderately elevated values, independently of age and sex. Further studies are needed for fast protocols that include resampling before 3 hours since biological variability is expected to be lower within such a short time frame. Pierre Douville receives a grant from Roche Diagnostics for a project unrelated to this report. References 1. de Lemos JA, Drazner MH, Omland T, et al.   Association of troponin T detected with a highly sensitive assay and cardiac structure and mortality risk in the general population. JAMA . 2010; 304: 2503- 2512. Google Scholar CrossRef Search ADS PubMed  2. Gore MO, Seliger SL, Defilippi CR, et al.   Age- and sex-dependent upper reference limits for the high-sensitivity cardiac troponin T assay. J Am Coll Cardiol . 2014; 63: 1441- 1448. Google Scholar CrossRef Search ADS PubMed  3. Eggers KM, Lind L, Venge P, et al.   Factors influencing the 99th percentile of cardiac troponin I evaluated in community-dwelling individuals at 70 and 75 years of age. Clin Chem . 2013; 59: 1068- 1073. Google Scholar CrossRef Search ADS PubMed  4. Biener M, Giannitsis E, Lamerz J, et al.   Prognostic value of elevated high-sensitivity cardiac troponin T levels in a low risk outpatient population with cardiovascular disease. Eur Heart J Acute Cardiovasc Care . 2016; 5: 409- 418. Google Scholar CrossRef Search ADS PubMed  5. Franzini M, Lorenzoni V, Masotti S, et al.   The calculation of the cardiac troponin T 99th percentile of the reference population is affected by age, gender, and population selection: a multicenter study in Italy. Clin Chim Acta . 2015; 438: 376- 381. Google Scholar CrossRef Search ADS PubMed  6. Chenevier-Gobeaux C, Bailleul S, Mzabi A, et al.   Upper reference limits of high-sensitivity cardiac troponin T in a general population: comparison with those of sensitive cardiac troponin I. Clin Lab . 2013; 59: 333- 336. Google Scholar CrossRef Search ADS PubMed  7. Roffi M, Patrono C, Collet JP, et al.  ; Management of Acute Coronary Syndromes in Patients Presenting Without Persistent ST-Segment Elevation of the European Society of Cardiology. 2015 ESC guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation: task force for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation of the European Society of Cardiology (ESC). Eur Heart J . 2016; 37: 267- 315. Google Scholar CrossRef Search ADS PubMed  8. McMullin N, Lindsell CJ, Lei L, et al.  ; EMCREG i*trACS Investigators. Outcomes associated with small changes in normal-range cardiac markers. Am J Emerg Med . 2011; 29: 162- 167. Google Scholar CrossRef Search ADS PubMed  9. Menacer S, Claessens YE, Meune C, et al.   Reference range values of troponin measured by sensitive assays in elderly patients without any cardiac signs/symptoms. Clin Chim Acta . 2013; 417: 45- 47. Google Scholar CrossRef Search ADS PubMed  10. Mansour M, Clark L, Scott T, et al.   Considerations for establishing a reference interval for elderly individuals in the emergency department with the high-sensitivity cardiac troponin T assay. Clin Chim Acta . 2013; 421: 85- 86. Google Scholar CrossRef Search ADS PubMed  11. Sajeev JK, New G, Roberts L, et al.   High sensitivity troponin: does the 50% delta change alter clinical outcomes in chest pain presentations to the emergency room? Int J Cardiol . 2015; 184: 170- 174. Google Scholar CrossRef Search ADS PubMed  12. Buccelletti F, Galiuto L, Marsiliani D, et al.   Comparison of diagnostic accuracy between three different rules of interpreting high sensitivity troponin T results. Intern Emerg Med . 2012; 7: 365- 370. Google Scholar CrossRef Search ADS PubMed  13. Frankenstein L, Wu AH, Hallermayer K, et al.   Biological variation and reference change value of high-sensitivity troponin T in healthy individuals during short and intermediate follow-up periods. Clin Chem . 2011; 57: 1068- 1071. Google Scholar CrossRef Search ADS PubMed  14. Vasile VC, Saenger AK, Kroning JM, et al.   Biological and analytical variability of a novel high-sensitivity cardiac troponin T assay. Clin Chem . 2010; 56: 1086- 1090. Google Scholar CrossRef Search ADS PubMed  15. Nordenskjöld AM, Ahlström H, Eggers KM, et al.   Short- and long-term individual variation in cardiac troponin in patients with stable coronary artery disease. Clin Chem . 2013; 59: 401- 409. Google Scholar CrossRef Search ADS PubMed  16. Reichlin T, Schindler C, Drexler B, et al.   One-hour rule-out and rule-in of acute myocardial infarction using high-sensitivity cardiac troponin T. Arch Intern Med . 2012; 172: 1211- 1218. Google Scholar CrossRef Search ADS PubMed  17. Biener M, Mueller M, Vafaie M, et al.   Comparison of a 3-hour versus a 6-hour sampling-protocol using high-sensitivity cardiac troponin T for rule-out and rule-in of non-STEMI in an unselected emergency department population. Int J Cardiol . 2013; 167: 1134- 1140. Google Scholar CrossRef Search ADS PubMed  18. Mueller M, Biener M, Vafaie M, et al.   Absolute and relative kinetic changes of high-sensitivity cardiac troponin T in acute coronary syndrome and in patients with increased troponin in the absence of acute coronary syndrome. Clin Chem . 2012; 58: 209- 218. Google Scholar CrossRef Search ADS PubMed  19. Reichlin T, Irfan A, Twerenbold R, et al.   Utility of absolute and relative changes in cardiac troponin concentrations in the early diagnosis of acute myocardial infarction. Circulation . 2011; 124: 136- 145. Google Scholar CrossRef Search ADS PubMed  20. Pretorius CJ, Wilgen U, Ungerer JP. Serial cardiac troponin differences measured on four contemporary analyzers: relative differences, actual differences and reference change values compared. Clin Chim Acta . 2012; 413: 1786- 1791. Google Scholar CrossRef Search ADS PubMed  21. Scharnhorst V, Krasznai K, van’t Veer M, et al.   Variation of cardiac troponin I and T measured with sensitive assays in emergency department patients with noncardiac chest pain. Clin Chem . 2012; 58: 1208- 1214. Google Scholar CrossRef Search ADS PubMed  22. Simpson AJ, Potter JM, Koerbin G, et al.   Use of observed within-person variation of cardiac troponin in emergency department patients for determination of biological variation and percentage and absolute reference change values. Clin Chem . 2014; 60: 848- 854. Google Scholar CrossRef Search ADS PubMed  23. Reichlin T, Cullen L, Parsonage WA, et al.   Two-hour algorithm for triage toward rule-out and rule-in of acute myocardial infarction using high-sensitivity cardiac troponin T. Am J Med . 2015; 128: 369- 379.e4. Google Scholar CrossRef Search ADS PubMed  © American Society for Clinical Pathology, 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

American Journal of Clinical PathologyOxford University Press

Published: Apr 28, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off