Clinical Chemistry 58:1 209–218 (2012)
Proteomics and Protein Markers
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 Matthias Mueller,1 Moritz Biener,1 Mehrshad Vafaie,1 Susanne Doerr,1 Till Keller,2 Stefan Blankenberg,2 Hugo A. Katus,1 and Evangelos Giannitsis1*
BACKGROUND: We evaluated kinetic changes of highsensitivity cardiac troponin T (hs-cTnT) in patients with acute coronary syndrome (ACS) and patients with hs-cTnT increases not due to ACS to rule in or rule out non–ST-segment elevation myocardial infarction (STEMI).
hs-cTnT was measured serially in consecutive patients presenting to the emergency department. Patients with ACS who had at least 2 hs-cTnT measurements within 6 h and non-ACS patients with hs-cTnT concentrations above the 99th percentile value (14 ng/L) were enrolled to compare absolute and relative kinetic changes of hs-cTnT.
or fall of at least 9.2 ng/L in the entire study population and 6.9 ng/L in selected ACS patients seems adequate to rule-out non-STEMI. However, ␦-values are useful to rule-in non-STEMI only in a specific ACS population. © 2011 American Association for Clinical Chemistry
METHODS:
For discrimination of non-STEMI (n ⫽ 165) in the entire study population (n ⫽ 784), the absolute ␦ change with the ROC-optimized value of 9.2 ng/L yielded an area under the curve of 0.898 and was superior to all relative ␦ changes (P ⬍ 0.0001). The positive predictive value for the absolute ␦ change was 48.7%, whereas the negative predictive value was 96.5%. In a specific ACS population with exclusion of STEMI (n ⫽ 342), the absolute ␦ change with the ROC-optimized value of 6.9 ng/L yielded a positive predictive value of 82.8% and a negative predictive value of 93.0%. In comparison to the ⱖ20% relative ␦ change, the ROCoptimized absolute ␦ change demonstrated a significantly added value for the entire study population and for the ACS cohort (net reclassification index 0.331 and 0.499, P ⬍ 0.0001). RESULTS:
Absolute ␦ changes appear superior to relative ␦ changes in discriminating non-STEMI. A rise
CONCLUSIONS:
1
Department of Internal Medicine III, Cardiology, University Hospital Heidelberg, Heidelberg, Germany; 2 Department of General and Interventional Cardiology, The University Heart Center at the University Medical Center HamburgEppendorf, Hamburg, Germany. * Address correspondence to this author at: Medizinische Klinik III, Im Neuenheimer Feld 410, 69120 Heidelberg, Germany. Fax ⫹49-6221-56-5516; e-mail
[email protected]. Received July 1, 2011; accepted October 31, 2011.
The Joint European Society of Cardiology/American College of Cardiology/American Heart Association/ World Heart Federation Task Force for the redefinition of acute myocardial infarction (AMI)3 has recommended that a diagnosis of AMI be made only in the presence of a rise and/or fall of cardiac troponin, with at least 1 value above the 99th percentile reference value (1 ). These findings in conjunction with a clinical context suggesting myocardial ischemia as the underlying mechanism indicate that a diagnosis of AMI should be made. Otherwise, other acute heart disease causing a dynamic cardiac troponin release should be considered (2 ). Recently, Javed et al. reported that fewer than one third of patients with increased cardiac troponin were found to have AMI (3 ). Therefore, knowledge about the magnitude of concentration changes (␦) in AMI in the absence of acute coronary syndrome (ACS) is essential to the definition of an optimal dynamic metric that allows discrimination of acute from a chronic conditions and of AMI from non–ACS-related conditions that cause cardiac troponin increases. Unfortunately, the magnitude at which the increase or decrease is indicative of an acute rather than a chronic cardiac troponin increase is unclear, and it is still debatable whether biological variation plays a role
Previously published online at DOI: 10.1373/clinchem.2011.171827 Nonstandard abbreviations: AMI, acute myocardial infarction; ACS, acute coronary syndrome; RCV, reference change value; ED, emergency department; hs-cTnT, high-sensitivity cardiac troponin T; STEMI, ST-segment elevation myocardial infarction; UAP, unstable angina pectoris; PCI, percutaneous coronary intervention; NPV, negative predictive value; PPV, positive predictive value; AUC, area under the curve; NRI, net reclassification index; IQR, interquartile range.
3
209
and whether absolute or relative ␦ change is the ideal metric for cardiac troponin measurement. The National Academy of Clinical Biochemistry has recommended a ␦ change in cardiac troponin of ⱖ20% 6 –9 h after presentation, but this recommendation is based on analytical considerations only in the subset of patients with end-stage renal disease or other conditions with low baseline-concentration increases of cardiac troponin (4 ). Several investigators have proposed higher relative ␦ changes between 30%–250% to increase the diagnostic specificity and thus improve diagnosis of AMI (5– 8 ). Recent study results suggest the use of reference change values (RCVs), which are the relative changes determined from biological variability of cardiac troponin (9 –11 ). Previous studies have investigated patients presenting with chest pain or more selected ACS populations and thus have not provided data on dynamic changes of cardiac troponin in other acute heart diseases for which patients commonly present to an emergency department (ED). Therefore, we evaluated dynamic ␦ changes of high-sensitivity cardiac troponin T (hs-cTnT) in patients with ACS and cardiac troponin increases in the absence of ACS who presented with acute symptoms to an ED. We determined the diagnostic accuracy of absolute and relative ␦ changes of cardiac troponin in discrimination of AMI. Methods During a 6-month period we measured hs-cTnT serially on presentation and at least after 3 or 6 h in all consecutive patients presenting to the internal medicine ED and chest pain unit of the University Hospital Heidelberg. Patients with ACS who received a second blood draw within 6 h and patients with non-ACS conditions with at least 1 hs-cTnT concentration above the 99th percentile value (14 ng/L) qualified for evaluation of relative or absolute ␦ changes within 3– 6 h after admission to rule in or rule out non–ST-segment elevation myocardial infarction (non-STEMI). AMI, without further differentiation into type I or type II, was diagnosed according to the criteria of the universal definition with detection of a rising and/or falling pattern of hs-cTnT and evidence of myocardial ischemia (1 ). Because the magnitude of the rise and/or fall for the diagnosis of AMI is still not established, we used a 20% rise and/or fall as a minimum to define an acute change, including all available hs-cTnT measurements within 24 h after the initial blood draw. In addition, an absolute concentration change of ⱖ5 ng/L between baseline and the highest consecutive value was required to diagnose AMI. We excluded patients with ST-segment elevations or new left bundle-branch block on presentation be210 Clinical Chemistry 58:1 (2012)
cause the diagnosis of STEMI is made by electrocardiogram and biomarker testing is not recommended for STEMI patients. A diagnosis of unstable angina pectoris (UAP) was made if ACS was suspected clinically but hs-cTnT concentrations were consistently below the 99th percentile value during serial sampling for at least 6 h. Moreover, patients with increased hs-cTnT and a typical presentation of ACS but without a relative ␦ change ⱖ20% or an absolute concentration difference of ⬍5 ng/L were found to have UAP with hs-cTnT increases due to underlying chronic cardiac damage or severe renal failure. In addition, the diagnosis of UAP required the presence of typical symptoms together with a history of coronary artery disease and previous coronary intervention or detection of a culprit lesion of ⱖ50% on coronary angiogram, or objective evidence of myocardial ischemia on stress testing. To discriminate procedure-related MI, we excluded patients who underwent percutaneous coronary intervention (PCI) and developed subsequent hscTnT increases after PCI before a final diagnosis of unstable angina or non-STEMI could be made. In contrast, we did not exclude patients with increased baseline hs-cTnT and declining values after PCI. In the absence of clinical variables suggestive of myocardial ischemia, increased hs-cTnT was interpreted as unrelated to AMI, and the underlying reason of myocardial damage was actively sought. Reasons for increased hs-cTnT in the absence of ACS were categorized into cardiac, extracardiac, and uncertain. Cardiac causes comprised acutely decompensated heart failure, decompensated valve disease, Tako-Tsubo cardiomyopathy, myocarditis, pulmonary embolism, and atrial or ventricular tachyarrhythmias. Extracardiac causes included severe kidney dysfunction/end-stage renal disease and sepsis. Final diagnosis of ACS or non-ACS was based on all available clinical, laboratory, and imaging findings adjudicated by an expert committee of 2 independent cardiologists blinded to the investigational biomarker results. A third cardiologist refereed in situations of disagreement. All medical decisions including the need and timing of coronary angiography, coronary intervention, or further diagnostic work-up were left to the discretion of the attending cardiologist. The study was performed according to the principles of the Declaration of Helsinki and approved by the local ethics committee. Written informed consent was obtained from all participating patients. Follow-up was accomplished via telephone contact or questionnaire at least 6 months after discharge. LABORATORY MEASUREMENTS
We measured cardiac troponin on COBAS E411 using the hs-cTnT assay (Roche Diagnostics), which is com-
Kinetics in Non-STEMI and Non-ACS cTnT
mercially available in Germany (not yet available in the US). The limit of the blank (3 ng/L) and limit of detection (5 ng/L) were determined in accordance with CLSI guideline EP17-A. The interassay CV was 8% at 10 ng/L and 2.5% at 100 ng/L. The intraassay CV was 5% at 10 ng/L and 1% at 100 ng/L (12 ). Reference interval values were established from a multicenter reference study and the 99th percentile value was determined at 14 ng/L (13 ). Along with admission values, several kinetic metrics were calculated from serial measurements including relative ␦ change [(Cmax(3 h or 6h) ⫺ Cbaseline)/Cbaseline ⫻ 100, where C is hs-cTnT concentration] given as percentage change in either direction (rise or fall) of hscTnT from baseline, absolute ␦ change between highest hs-cTnT concentration from 3- or 6-h sample and baseline [Cmax 3h or 6h ⫺ Cbaseline] expressed as nanograms per liter, and peak hs-cTnT concentration within 6 h. STATISTICAL ANALYSIS
Continuous variables were tested for normal distribution and were presented either as mean (SD) or as medians with 25th and 75th percentiles. We compared groups by using the 2 test for categorical variables and ANOVA for continuous variables. We determined optimal thresholds for relative and absolute ␦ changes from ROC curves on the basis of the continuously measured biomarker concentrations. We calculated diagnostic sensitivities, diagnostic specificities, negative predictive values (NPVs), and positive predictive values (PPVs) for relative and absolute ␦ changes for classification of final diagnosis of non-STEMI. To compare areas under the curve (AUCs) we used the test of DeLong et al. (14 ). To demonstrate an added value of other kinetic changes compared to a 20% relative change, we determined net reclassification index (NRI) according to the method by Pencina et al. (15 ). All analyses on performance were executed for the entire population and in 4 categories of incremental baseline hs-cTnT intervals to adjust for the effects of absolute baseline concentrations. The categories included baseline hs-cTnT concentrations of ⬍14 ng/L, 14 – 49 ng/L, 50 –99 ng/L, and ⱖ100 ng/L. SPSS 15.0 and MedCalc 11.1 statistical software packages were used. All tests were 2-tailed, and a P value ⬍0.05 was considered statistically significant. Results BASELINE
During a 6-month recruitment period a total of 863 patients qualified for the kinetic study. Non-STEMI was diagnosed in 165 patients (19.1%) and UAP in 177 patients (20.5%). In 442 patients (51.2%) cardiac tro-
ponin increases were not due to ACS (non-ACS conditions). We excluded 64 patients (7.4%) presenting with ST-segment elevations or new left bundle-branch block on electrocardiogram. Moreover, we excluded 15 patients (1.7%) who underwent PCI and developed increasing hs-cTnT values after PCI but before the final diagnosis was made. Thus, the entire study population consisted of 784 patients, including patients with ACS and non-ACS conditions. The ACS study population consisted of 342 patients with non-STEMI and UAP. Reasons for cardiac troponin increases in non-ACS conditions comprised cardiac diseases in 152 (34.4%), extracardiac diseases in 159 (36.0%), and uncertain causes in 131 patients (29.6%). The baseline characteristics of the entire study population (n ⫽ 784) subdivided by final diagnosis are displayed in Table 1. As expected, patients with non-STEMI and patients with hs-cTnT increases not due to ACS differed with respect to most baseline demographic characteristics and laboratory and angiographic findings. In addition, both groups within the ACS spectrum showed significant differences, mainly in angiographic baseline characteristics. More details on baseline characteristics are provided in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol58/issue1. SERIAL CHANGES
The median number of hs-cTnT measurements per individual was 3 (25th to 75th percentile: 2–3). Table 2 demonstrates hs-cTnT concentrations on presentation and during consecutive sampling, as well as maximal absolute and relative ␦ changes. Fig. 1 shows the maximal individual relative and absolute ␦ changes within 6 h for patients with non-STEMI, UAP, or non-ACS conditions. All 165 patients with non-STEMI and non-ACS conditions as well as 91 patients with UAP (51.4%) had hs-cTnT concentrations ⱖ99th percentile in a least 1 sample during the initial 6 h. Within this time frame, 62.4% of patients with non-ACS conditions (n ⫽ 276) and 75.2% of patients with non-STEMI (n ⫽ 124) fulfilled the 20% ␦ change criterion. Conversely, 24.8% of patients with a final diagnosis of non-STEMI (n ⫽ 41) did not fulfill this diagnostic criterion within the initial 6 h, but did spontaneously during subsequent sampling within 24 h. Compared to non-STEMI patients with a relative ␦ change ⱖ20%, those patients with a ␦ change ⬍20% presented later after onset of symptoms, at 20 h [interquartile range (IQR) 12– 48 h] vs 12 h (IQR 1– 48 h) (P ⫽ 0.04), and showed higher hs-cTnT on presentation [306.4 ng/L (IQR 131.7–589.2) vs 45.7 ng/L (IQR 20.9 –201.4), P ⬍ 0.001]. Clinical Chemistry 58:1 (2012) 211
Table 1. Baseline demographic, laboratory, and angiographic characteristics of hs-cTnT elevation in non-STEMI, unstable angina, and non-ACS patients. Non-STEMI (n ⴝ 165)
Age, mean (range), years Age ⬎75 years, n (%) Male sex, n (%) NT-proBNP,b mean (range), ng/L eGFR, mean (SD) or (range) eGFR ⬍60, n (%)
70.4 (59.1–77.5)
UAP (n ⴝ 177)
Non-ACS (n ⴝ 442)
71.4 (61.8–78.7)
74.1 (66.6–81.2)
52 (31.5%)
64 (39.0%)
205 (46.4%)
121 (73.3%)
106 (59.9%)
173 (39.1%)a
559.5 (209.0–1294.0) 73.2 (30.7)
594.0 (200–3629)
2943 (994.5–7982.5)a 63.0 (43.3–80.6)c
69.1 (24.3)
58 (35.2%)
58 (32.8%)
202 (45.7%)a
113 (68.5%)
133 (75.1%)
58 (13.1%)a
Dyspnea
13 (7.9%)
7 (4.0%)
149 (33.7%)d
Syncope
2 (1.2%)
Leading symptom, n (%) Angina pectoris
Other
37 (22.4%)
29 (6.6%)c
0 37 (20.9%)
206 (46.6%)
History, n (%) CHF
28 (17.0%)
43 (24.3%)
88 (20.0%)
CAD
151 (91.5%)
149 (84.2%)
254 (57.5%)a
PCI
63 (38.2%)
99 (55.9%)
121 (27.4%)
CABG
21 (12.7%)
23 (13.0%)
44 (10.0%)
PAD
18 (10.9%)
24 (13.6%)
38 (8.6%)
Stroke
12 (7.3%)
13 (7.3%)
33 (7.5%)
COPD
19 (11.5%)
21 (11.9%)
81 (18.3%) 133 (30.1%)
Risk factors, n (%) Diabetes mellitus
58 (35.2%)
65 (36.7%)
Cholesterolemia
105 (63.6%)
138 (78.0%)
228 (51.6%)
Hypertension
139 (84.2%)
166 (93.8%)d
351 (79.4%)
Active smoking
33 (20.0%)
20 (11.3%)d
39 (8.8%)
Ex-smoker
54 (32.7%)
60 (33.9%)
146 (33.0%)
Family history GRACE score, mean (SD) Coronary angiography, n (%) Coronary artery disease, n (%)
14 (8.5%) 129.0 (33.8)
14 (7.9%)
61 (13.9%) 140.6 (32.7)c
119.3 (31.0)
147 (89.1%)
127 (71.8%)c
129 (29.2%)a
146 (88.5%)
c
109 (24.7%)a
124 (70.1%)
0 VD
2 (1.2%)
9 (5.1%)
55 (12.4%)
1 VD
23 (13.9%)
14 (7.9%)
13 (2.9%)
2 VD
22 (13.3%)
28 (19.4%)
3 VD Left main trunk PCI
94 (57.0%) 5 (3.0%) 105 (63.6%)
c
76 (42.9%) 0
67 (37.9%)c
16 (3.6%) 44 (10.0%)a 1 (0.2%) 13 (2.9%)a
P ⬍0.0001 vs non-STEMI. NT-proBNP, N-terminal pro-B type natriuretic peptide; eGFR, estimated glomerular filtration rate; CHF, chronic heart failure; CAD, coronary artery disease; CABG, coronary artery bypass graft; PAD, peripheral artery disease; COPD, chronic obstructive pulmonary disease; VD, vessel disease. c P ⬍0.01 vs non-STEMI. d P ⬍0.05 vs non-STEMI. a
b
PERFORMANCE OF SERIAL ␦ CHANGES FOR RULE-IN AND RULE-OUT OF NON-STEMI
In the entire study population (n ⫽ 784), AUCs were determined from ROC analysis for continuous values 212 Clinical Chemistry 58:1 (2012)
including baseline hs-cTnT concentrations and maximum hs-cTnT concentrations within 6 h as well as relative and absolute ␦ changes. Comparison of AUC values demonstrated a significantly better performance of
Kinetics in Non-STEMI and Non-ACS cTnT
Table 2. Baseline hs-cTnT, maximum absolute and relative ␦ changes. Serial measurement
Non-ACS (n ⴝ 442)
5 (4–6)
4 (3–4)
3 (2–4)a
No. of samples 0–6 h
3 (2–3)
3 (3–3)
2 (2–3)
No. of samples 24–72 h
1 (1–1)
1 (0–1)
a
0 (0–1)a a
133 (80.1%)
390 (88.2%)b
168 (94.9%)
Sample at 6 h
143 (86.7%)
142 (80.2%)
365 (82.6%)
Any sample at 24–72 h
145 (87.9%)
104 (58.8%)a
207 (46.8%)a
ⱖ99th percentile 0 h
159 (96.4%)
89 (50.3%)a
418 (94.6%)
ⱖ99th percentile 3 h
121 (92.4%)
a
79 (47.0%)
344 (88.2%)
ⱖ99th percentile 6 h
132 (92.3%)
68 (47.9%)a
233 (63.8%)a
165 (100%)
a
ⱖ99th percentile 0–6 h hsTnT 0 h (baseline), ng/L hsTnT 3 h, ng/L hsTnT 6 h, ng/L
91 (51.4%)
442 (100%)
92.0 (27.3–312.3)
14.0 (5.1–26.8)a
31.5 (19.5–60.4)a
130.3 (46.4–378.5)
11.6 (5.0–25.5)a
31.3 (19.4–62.1)a
188.8 (62.4–555.1)
a
31.4 (19.3–63.6)a
a
49.2 (27.6–91.6)a
10.8 (3.8–25.3)
hsTnT 24 h, ng/L
341.1 (114.1–871.0)
18.3 (6.5–59.1)
hsTnT 48 h, ng/L
398.2 (116.1–907.2)
54.0 (20.7–134.6)a
61.8 (32.4–184.9)a
a
92.0 (31.6–258.6)a
hsTnT 72 h, ng/L Early peak (0–6 h), ng/L Late peak (24–72 h), ng/L Relative change, baseline to 6 h, % Absolute change, baseline to 6 h ng/L b
UAP (n ⴝ 177)
No. of samples total
Sample at 3 h
a
Non-STEMI (n ⴝ 165)
410.1 (152.7–1315.0) 245.5 (74.5–642.0)
68.7 (20.7–197.1) a
35.1 (22.1–69.3)a
16.2 (6.2–31.4)
381.0 (124.2–1148.2)
24.4 (11.5–83.5)
52.3 (28.1–101.9)a
53.9 (20.0–200.3)
14.2 (7.9–22.4)a
15.6 (8.1–28.8)a
56.5 (18.2–250.0)
a
a
2.8 (1.1–5.6)
5.1 (2.6–11.3)a
P ⱕ 0.0001 vs non-STEMI. P ⬍0.05 vs non-STEMI.
absolute ␦ changes (AUC ⫽ 0.898) than relative ␦ changes (AUC ⫽ 0.752, P ⬍ 0.0001), baseline hs-cTnT (AUC ⫽ 0.731, P ⬍ 0.0001), and maximum hs-cTnT concentrations within 6 h (AUC ⫽ 0.830, P ⬍ 0.0001) (Fig. 2A). The diagnostic sensitivities, diagnostic specificities, PPVs and NPVs derived from ROC-based optimal cutoff values for relative and absolute ␦ changes as well as for other relevant ␦ changes are displayed in Table 3. The ROC-based optimal cutoff value of ⱖ9.2 ng/L for absolute ␦ change showed a diagnostic sensitivity of 89.7% (95% CI 84.0 –93.9) with a NPV of 96.5% (95% CI 94.4 –97.9). Diagnostic specificities were lower, yielding PPVs of 48.7% (95% CI 42.9 –54.5) for the absolute ␦ change of 9.2 ng/L. Compared to a ⱖ20% relative ␦ change, the absolute ␦ change of ⱖ5 ng/L (NRI ⫽ 0.185, P ⬍ 0.001), ⱖ20 ng/L (NRI ⫽ 0.272, P ⬍ 0.001), and the ROC-optimized ␦ change of 9.2 ng/L (NRI ⫽ 0.311, P ⬍ 0.0001) demonstrated a significant added value. Analysis taking into consideration baseline hscTnT concentrations demonstrated a better performance of absolute changes than relative ␦ changes at baseline hs-cTnT concentrations ⬍14 ng/L (n ⫽ 118
patients) with an ROC-based optimal cutoff for absolute changes of 4.7 ng/L (AUC ⫽ 0.96 vs AUC ⫽ 0.815, P ⫽ 0.013) and baseline concentrations of ⱖ100 ng/L (n ⫽ 155) with a cutoff value for absolute change of 47.6 ng/L (AUC ⫽ 0.777 vs 0.695, P ⫽ 0.0047). At baseline concentrations of 14 – 49 ng/L (n ⫽ 404) the relative ␦ change was significantly superior to absolute ␦ change in identifying patients with non-STEMI using a cutoff value of 43.5% for relative ␦ changes (AUC ⫽ 0.928 vs 0.902, P ⫽ 0.009). In the baseline concentration range of 50 –99 ng/L (n ⫽ 107) no significant difference was found between absolute and relative ␦ changes (AUC ⫽ 0.81 vs 0.799, P ⫽ 0.521) (Fig. 3). By comparing performance of relative and absolute ␦ changes related to time from symptom onset to admission, we observed an inferior performance of relative ␦ changes among patients presenting later than 4 h after onset of symptoms. In contrast, performance of absolute ␦ changes was superior to relative ␦ changes independent of time to symptom onset. Moreover, no significant difference was found for performance of absolute ␦ changes related to the time from symptom onset to admission (see online Supplemental Fig. 1). The diagnostic performance of different ␦ changes with Clinical Chemistry 58:1 (2012) 213
Fig. 1. Distribution of absolute (A) and relative (B) ␦ change in patients with a final diagnosis of non-STEMI (n ⴝ 165) or UAP (n ⴝ 177) and in patients with increased hs-cTnT due to non-ACS conditions (n ⴝ 442). Compared to UAP and non-ACS conditions, patients with non-STEMI showed significantly increased absolute ␦ changes (P ⬍ 0.0001, respectively) and relative ␦ changes (P ⬍ 0.0001, respectively).
addition of the 99th percentile cutoff value for hs-cTnT are shown in online Supplemental Table 1. In a specific ACS population (n ⫽ 342), comparison of AUC values demonstrated a significantly better performance of absolute (AUC ⫽ 0.941) than relative ␦ changes (AUC ⫽ 0.741, P ⬍ 0.0001) as well as baseline (AUC ⫽ 0.836, P ⬍ 0.0001) and peak hs-cTnT concentrations (AUC ⫽ 0.894, P ⬍ 0.0001) (Fig. 2B). A value of 6.9 ng/L was detected as the ROC-optimized absolute ␦ change. The performance of ␦ changes for rule-in and rule-out of non-STEMI in the specific ACS population is shown in online Supplemental Data Table 2.
Discussion Our findings demonstrate the superiority of absolute ␦ changes in identification of non-STEMI within 3– 6 h in a population of consecutive patients presenting to an ED with ACS and troponin increases due to non-ACS conditions. Concentration changes of hs-cTnT were significantly higher in non-STEMI than in other acute cardiac or extracardiac disease and there was a wide overlap of ␦ values, particularly relative ␦ changes. In direct comparison of kinetic changes, absolute ␦ changes outperformed relative ␦ changes (AUC ⫽
Fig. 2. Comparison of areas under the curve of baseline hs-cTnT, peak hs-cTnT within 6 hours, relative ␦ change and absolute ␦ change for prediction of non-STEMI in the entire study population (A) and ACS population (B).
214 Clinical Chemistry 58:1 (2012)
Kinetics in Non-STEMI and Non-ACS cTnT
Table 3. Performance of kinetic changes in hs-cTnT within the initial 6 h for rule-in and rule-out of non-STEMI in the entire study population. NRIa
Sensitivity (95% CI)
Specificity (95% CI)
PPV (95% CI)
NPV (95% CI)
Change ⱖ20%
75.2 (67.8–81.5)
58.1 (54.1–62.0)
32.4 (27.7–37.3)
89.8 (86.4–92.5)
—
Change ⱖ30%
63.6 (55.8–71.0)
75.1 (71.5–78.5)
40.5 (34.5–46.8)
88.6 (85.5–91.2)
0.055
Change ⱖ39.8%b
57.6 (49.7–65.2)
83.0 (79.8–85.9)
47.5 (40.4–54.6)
88.0 (85.1–90.5)
0.072
Change ⱖ50%
52.7 (44.8–60.5)
87.5 (84.7–90.0)
53.1 (45.1–60.9)
87.4 (84.5–89.9)
0.070
Change ⱖ100%
35.4 (28.1–43.2)
95.3 (93.3–96.8)
66.7 (55.8–76.4)
84.8 (81.9–87.3)
⫺0.029
Change ⱖ250%
23.2 (17.0–30.4)
99.2 (98.1–99.7)
88.4 (74.9–96.1)
83.0 (80.0–85.6)
⫺0.110
Change ⱖ5.0 ng/L
100 (97.8–100)
55.4 (51.4–59.4)
37.4 (32.9–42.1)
98.3 (96.3–99.4)
0.185c
Change ⱖ9.2 ng/Lb
89.7 (84.0–93.9)
74.8 (71.2–78.2)
48.7 (42.9–54.5)
96.5 (94.4–97.9)
0.311c
Change ⱖ20 ng/L
72.7 (65.2–79.4)
87.7 (84.9–90.2)
61.2 (54.0–68.1)
92.4 (89.9–94.4)
0.272c
Change ⱖ50 ng/L
52.1 (44.2–60.0)
95.2 (93.2–96.7)
74.1 (65.2–81.8)
88.2 (85.5–90.5)
0.140
Change ⱖ100 ng/L
37.6 (30.2–45.4)
97.4 (95.8–98.5)
79.5 (68.8–87.8)
85.4 (82.6–87.9)
0.017
Change ⱖ200 ng/L
28.5 (21.7–36.0)
98.7 (97.5–99.4)
85.5 (73.3–93.5)
83.8 (80.9–86.4)
⫺0.061
Vs relative ␦ change ⱖ20%. ROC-based optimal cutoff for discrimination of non-STEMI. c P ⬍ 0.001. a
b
0.898 vs 0.752, P ⬍ 0.0001). Compared to a ⱖ20% relative ␦ change, the ROC-optimized absolute ␦ change of 9.2 ng/L for the entire study population and 6.9 ng/L for a specific ACS population demonstrated a significant added value (NRI ⫽ 0.331 and 0.499, P ⬍ 0.0001, respectively). In addition, the performance of absolute ␦ changes was independent of time from symptom onset to admission, whereas the performance of relative ␦ changes declined 4 h after symptom
Fig. 3. Performance of absolute and relative ␦ change related to 4 baseline hs-cTnT categories, i.e. 100 ng/L.
onset. Because 24.8% of all non-STEMI patients presented with relative ␦ changes ⬍20% within the initial 6 h, and these patients were more likely to have increased baseline hs-cTnT concentrations with prolonged time from symptom onset to admission, the conclusion can be made that this group of patients may already have reached a plateau of the cardiac troponin release curve. In contrast, only 10.3% of the nonSTEMI patients using the absolute ␦ change of 9.2 ng/L were below this cutoff in the initial 6 h. This result may indicate that absolute ␦ changes are also more diagnostically sensitive in the plateau phase of cardiac troponin release. More importantly, we found that the ROCoptimized absolute ␦ change was useful to rule out non-STEMI with a NPV of 96.5% in the entire study population and a NPV of 93.0% in patients with ACS. In comparison, the use of the ROC-optimized relative ␦ changes for rule out yielded NPVs of 88.0% for the entire study population. However, with only weak PPVs for the ROC-optimized kinetic change values of 48.7% for absolute and 47.5% for relative ␦ change, rule-in of non-STEMI in a population consisting of ACS and non-ACS-related troponin increases is extremely difficult. This finding can likely be explained by the relative overlap of kinetic changes in this cohort, thus decreasing their usefulness to rule in non-STEMI in such a population. In contrast, implementation of dynamic changes only to patients with ACS allows both rule-in and rule-out of non-STEMI with a NPV of 93.0%, and a PPV of 82.8% for the ROC-optimized Clinical Chemistry 58:1 (2012) 215
absolute change. Accordingly, it appears that the most appropriate strategy to differentiate AMI in a cohort with a high prevalence of acute hs-cTnT increases not due to ACS is rule-out using absolute concentration changes, whereas rule-in of AMI may not be achieved adequately in this cohort. In addition, our findings indicate a problem that not only the magnitude of rise and/or fall but also the time interval in which the kinetic change should be fulfilled is poorly defined. The data suggest that limiting the serial measurement to only 6 h would increase the chance of misclassifying patients with hs-cTnT increases within the initial 6 h but without kinetic changes as non–ACS related conditions instead of nonSTEMI, particularly in situations with delayed presentation after onset of symptoms. This issue must be addressed in forthcoming trials. However, this does not affect patients without hs-cTnT increases in the first 3– 6 h since early rule-out can be accomplished reliably in these patients. PREVIOUS FINDINGS
Previously, Apple et al. tested the utility of percentage changes in cTnI of ⱖ10, ⱖ20, and ⱖ30%, and reported that ⱖ30% change in cTnI should be used as the optimal change in addition to either the baseline or follow-up concentration to improve specificity in patients presenting with symptoms of ACS (5 ). Our group demonstrated ROC-optimized relative ␦-change values for hs-cTnT between 117% and 243% within 3 and 6 h, respectively, for diagnosis of AMI in a selected small cohort of patients with evolving AMI (6 ). Conversely, Eggers et al. tested the impact of ␦-change values of ⱖ20%, ⱖ50%, and ⱖ100% and found that a change ⱖ50% would have resulted in more frequent falsenegative results in patients with an index diagnosis of AMI (7 ). Data on the impact of absolute ␦ changes on diagnostic performance are sparse. Hitherto, only 1 study has analyzed the diagnostic role of absolute ␦ changes in the diagnosis of AMI. Reichlin et al. tested the utility of absolute and relative ␦ changes for early diagnosis of AMI and also found that the performance of absolute ␦ changes was significantly superior to the performance of relative ␦ changes (16 ). The optimal cutoff value for absolute ␦ changes from baseline to 2-h follow-up sample was 7.0 ng/L, which is close to our proposed cutoff value of 9.2 ng/L from baseline to 3– 6 h. However, in contrast to the study by Reichlin et al. (16 ), who evaluated kinetic changes among patients with symptoms suggestive of AMI, we selected consecutive patients presenting to an ED with ACS or troponin increases in the absence of ACS. Therefore our study population showed a higher prevalence of patients with cardiac troponin increases at baseline (666 of 784 patients, 84.9%), in contrast to 35% of patients 216 Clinical Chemistry 58:1 (2012)
in the study by Reichlin et al. The high prevalence of patients with non–ACS-related cardiac troponin increases in relation to AMI leads to increasing difficulties in diagnosing AMI without kinetic changes. Moreover, in our study we showed a better diagnostic performance of the ROC-optimized absolute ␦ change of 9.2 ng/L in direct comparison to the ROC-optimized relative ␦ change of 39.8%. In contrast to Reichlin et al. (16 ) we could not demonstrate the diagnostic superiority of absolute ␦ changes independent of the underlying baseline concentration. Our data indicate that absolute ␦ changes were superior only in patients with low and high baseline concentrations and in the cohort that includes all baseline concentrations. However, in the area of baseline concentrations slightly above the 99thpercentile, relative ␦ changes showed a higher diagnostic accuracy compared to absolute ␦ changes. Therefore the use of relative ␦ change may be more appropriate for baseline hs-cTnT concentrations slightly above the 99thpercentile cutoff than in low or high hs-cTnT baseline concentrations because relative ␦ changes tend to under- or overestimate kinetic changes in these concentration ranges. More recently, data on biological variability of cardiac troponin in healthy individuals assessed by RCV suggested that biological variation may be more important for interpreting minor cardiac troponin concentration increases at or just above the 99thpercentile limit when assays of very high sensitivity are used (9– 11 ). For concentrations above the 99thpercentile up to 49 ng/L our data support the usefulness of relative ␦ changes of 43.5%, which is in the range of what has recently been reported for biological short-term variability (9, 11 ). Moreover, we and Reichlin et al. proved the feasibility of even small absolute ␦ changes during serial sampling by showing very high diagnostic accuracy. However, an important obstacle with biological variation is the fact that RCVs have to be calculated for every cardiac troponin assay separately, and RCVs strongly depend on the selection of the reference population. In addition, biological variation has not yet been evaluated for discrimination of non-STEMI against non– ACS-related troponin elevations. Thus, it appears that biological variation may represent a useful metric to discriminate acute from chronic cardiac troponin increases but need further prospective validation. LIMITATIONS
Because the reference standard in our study to diagnose non-STEMI was based on the 20% ␦ change within 24 h and not all non-STEMI patients fulfilled the 20% ␦ change within 6 h, we cannot exclude some bias when evaluating the performance of different absolute and relative ␦ changes. Moreover, not all patients in our
Kinetics in Non-STEMI and Non-ACS cTnT
study received angiography and had blood samples available exceeding 6 h. This potentially may create a bias for classification of non-STEMI and UAP. However, sampling time was at least 24 h in two thirds of all patients and the angiography rates are in compliance with other ED population studies or contemporary clinical trials (3, 17 ). Given ambiguous clinical presentation, we also cannot exclude that some patients whose illness would have qualified for type 2 MI were classified as non-ACS and vice versa. In addition, when using any absolute or relative ␦ changes, one has to be aware of the situation, that there may be some confounding due to biological variation. Finally, different prevalence of disease in different clinical settings may require other cutoff values for diagnosis of AMI in other populations. Therefore our results from a single observational study have to be confirmed by larger clinical trials. Conclusions The use of high-sensitivity cardiac troponin assays revealed that non-STEMI as well as other acute cardiac diseases demonstrated a considerable rise and/or fall of cardiac troponin that may easily exceed 20% with a substantial overlap in non-STEMI and other acute non-ACS conditions. These results explain why relative ␦ changes fail to rule in non-STEMI in a population consisting of ACS patients and patients with acute non–ACS-related troponin increases. Conversely, ␦-change criteria proved useful for rule-out of nonSTEMI, both in the entire study population and in the ACS cohort. Overall diagnostic performance of absolute ␦ changes was better than performance of relative changes owing to a higher specificity of absolute ␦
changes. With the use of absolute ␦ changes, a rise and/or fall of at least 9.2 ng/L for a population consisting of patients with ACS and non-ACS conditions, or 6.9 ng/L for an ACS population seems to be more adequate than relative ␦ changes for ruling out AMI.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article. Authors’ Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the Disclosures of Potential Conflict of Interest form. Potential conflicts of interest: Employment or Leadership: S. Blankenberg, guest editor, Clinical Chemistry, AACC. Consultant or Advisory Role: S. Blankenberg, Thermo Fisher; E. Giannitsis, Roche Diagnostics and BRAHMS. Stock Ownership: None declared. Honoraria S. Blankenberg, Roche Diagnostics, Abbott Diagnostics, Siemens, and Thermo Fisher; H.A. Katus, Novartis, Roche, and Bayer; E. Giannitsis, Roche Diagnostics, Siemens Healthcare, BRAHMS Biomarkers, and Mitsubishi Chemicals. Research Funding: E. Giannitsis, Roche Diagnostics Ltd, Switzerland; Mitsubishi Chemicals, Germany; Siemens Healthcare; BRAHMS Biomarkers, Clinical Diagnostics Division; and Thermo Fisher Scientific, Germany. Expert Testimony: None declared. Other: H.A. Katus has developed the cardiac troponin T assay and holds a patent jointly with Roche Diagnostics. Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript. Acknowledgments: We would like to thank Francisco OjedaEchevarria for help with statistical analysis.
References 1. Thygesen K, Alpert JS, White HD, Jaffe AS, Apple FS, Galvani M, et al. Universal definition of myocardial infarction. Circulation 2007;116: 2634 –53. 2. Agewall S, Giannitsis E, Jernberg T, Katus H. Troponin elevation in coronary vs. non-coronary disease. Eur Heart J 2011;32:404 –11. 3. Javed U, Aftab W, Ambrose JA, Wessel RJ, Mouanoutoua M, Huang G, et al. Frequency of elevated troponin I and diagnosis of acute myocardial infarction. Am J Cardiol 2009;104:9 –13. 4. Wu AH, Jaffe AS, Apple FS, Jesse RL, Francis GL, Morrow DA, et al. National Academy of Clinical Biochemistry laboratory medicine practice guidelines: use of cardiac troponin and B-type natriuretic peptide or N-terminal proB-type natriuretic peptide for etiologies other than acute coronary syndromes and heart failure. Clin Chem 2007;53:2086 –96. 5. Apple FS, Pearce LA, Smith SW, Kaczmarek JM, Murakami MM. Role of monitoring changes in sensitive cardiac troponin I assay results for early
6.
7.
8.
9.
diagnosis of myocardial infarction and prediction of risk of adverse events. Clin Chem 2009;55: 930 –7. Giannitsis E, Becker M, Kurz K, Hess G, Zdunek D, Katus HA. High-sensitivity cardiac troponin T for early prediction of evolving non-ST-segment elevation myocardial infarction in patients with suspected acute coronary syndrome and negative troponin results on admission. Clin Chem 2010; 56:642–50. Eggers KM, Jaffe AS, Venge P, Lindahl B. Clinical implications of the change of cardiac troponin I levels in patients with acute chest pain - an evaluation with respect to the Universal Definition of Myocardial Infarction. Clin Chim Acta 2011;412:91–7. Kavsak PA, Ko DT, Wang X, MacRae AR, Jaffe AS. 2007 universal myocardial infarction definition change criteria for risk stratification by use of a high-sensitivity cardiac troponin I assay. Clin Chem 2010;56:487–9. Wu AH, Lu QA, Todd J, Moecks J, Wians F.
10.
11.
12.
13.
Shortand long-term biological variation in cardiac troponin I measured with a high-sensitivity assay: implications for clinical practice. Clin Chem 2009; 55:52– 8. Vasile VC, Saenger AK, Kroning JM, Jaffe AS. Biological and analytical variability of a novel high-sensitivity cardiac troponin T assay. Clin Chem 2010;56:1086 –90. Frankenstein L, Wu AH, Hallermayer K, Wians FH Jr, Giannitsis E, Katus HA. 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 –71. Giannitsis E, Kurz K, Hallermayer K, Jarausch J, Jaffe AS, Katus HA. Analytical validation of a high-sensitivity cardiac troponin T assay. Clin Chem 2010;56:254 – 61. Saenger AK, Beyrau R, Braun S, Cooray R, Dolci A, Freidank H, et al. Multicenter analytical evaluation of a high-sensitivity troponin T assay. Clin Chim Acta 2011;412:748 –54.
Clinical Chemistry 58:1 (2012) 217
14. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44: 837– 45. 15. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS. Evaluating the added predictive ability
218 Clinical Chemistry 58:1 (2012)
of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med. 2008; 27:157–72; discussion 207–12. 16. Reichlin T, Irfan A, Twerenbold R, Reiter M, Hochholzer W, Burkhalter H, et al. Utility of absolute and relative changes in cardiac troponin concentrations in the early diagnosis of acute myocardial
infarction. Circulation 2011;124:136 – 45. 17. Cannon CP, Harrington RA, James S, Ardissino D, Becker RC, Emanuelsson H, et al. Comparison of ticagrelor with clopidogrel in patients with a planned invasive strategy for acute coronary syndromes (PLATO): a randomised double-blind study. Lancet 2010;375:283–93.
