- This study examined data on 624 patients from the ROMICAT or ROMICAT-II clinical trials who had blood samples tested with three high-sensitivity cardiac troponin (hs-cTn) assays because of suspected acute coronary syndrome (ACS)
- Only 384 of 1,027 samples were classified into the same analytical benchmark category (less than the limit of detection, limit of detection to 99th percentile or over the 99th percentile) and there was low concordance within benchmarks
- Among 242 patients who had serial hs-cTn measurements, concordance across the assays was only 75% for rule out, 16% for observation and 39% for admission
- Among patients for whom serial hs-cTn suggested rule-out, 19.6% to 21.1% had ACS according to computed tomographic angiography or single-photon emission computed tomography
- Clinicians should be aware that differences between hs-cTn assays may translate into substantial differences between hospitals in patient care (referral to further diagnostic testing and detection of underlying coronary artery disease)
Three high-sensitivity cardiac troponin (hs-cTn) assays are now FDA-approved for the diagnosis of acute coronary syndrome (ACS). The cutpoints used to make patient management decisions are specific to each assay.
In multiple studies, all of these assays have demonstrated high negative predictive values and high specificities. However, in a substudy of two clinical trials, Júlia Karády, MD, an instructor in the Department of Radiology and a researcher in the Cardiovascular Imaging Research Center (CIRC) at Massachusetts General Hospital, and Udo Hoffmann, MD, MPH, professor of Radiology and director of the CIRC, and colleagues determined for the first time that there are substantial differences between hs-cTn assays approved by the FDA in how they stratify individual blood samples and patients.
As the team reports in the Journal of the American College of Cardiology, this discordance can lead to different management recommendations for patients with suspected ACS who had an inconclusive evaluation in the emergency department.
The researchers examined data on 624 patients (1,027 blood samples) from the Rule Out Myocardial Infarction/Ischemia Using Computer Assisted Tomography (ROMICAT) trial, published in the Journal of the American College of Cardiology, and the ROMICAT-II trial, methodology published in the American Heart Journal.
Patients had suspected myocardial infarction or unstable angina but were referred for further noninvasive testing because conventional troponin measurement was negative and electrocardiography did not suggest ischemia. Patients included in the current study had blood samples tested with all three commercially available hs-cTn assays:
- Roche Elecsys 2010
- Abbott ARCHITECT i2000SR
- Siemens HsVista
The FDA suggests three analytic benchmarks for hs-cTn assays: less than the limit of detection (LOD), the limit of detection to 99th percentile or greater than 99th percentile. Only 37% of blood samples were classified into the same analytical benchmark category by all three assays.
Furthermore, there was low concordance across individual benchmarks:
- Less than the LOD: 11% of samples were concordant
- LOD to 99th percentile: 29%
- Greater than 99th percentile: 44%
242 patients from ROMICAT-II had serial hs-cTn measurements. Based on the difference between results at presentation and two hours later, concordance across the assays was:
- 75% for rule-out
- 16% for observation
- 39% for rule-in
hs-cTn and Imaging
148 of the 242 patients underwent computed tomographic angiography or single-photon emission computed tomography. Among the patients for whom serial hs-cTn suggested rule-out, 19.6% to 21.1% had obstructive coronary artery disease or myocardial ischemia according to imaging.
Implications for Clinicians and Professional Societies
Clinicians should be aware that differences between hs-cTn assays may translate into substantial differences between hospitals in patient care (referral to further diagnostic testing and detection of underlying coronary artery disease).
All the approved hs-cTn assays have excellent diagnostic properties, so the discrepancies found here are unlikely to be attributable to differences in performance. What's more plausible is that the discordances relate to differences in the age and gender distribution of the reference populations used when the assays were developed.
One solution might be to derive a 99th-percentile threshold for all hs-cTn assays using a standardized reference population. Another would be to develop methods for expressing the likelihood of ACS in a probabilistic fashion.
Learn more about the Cardiovascular Imaging Research Center
Learn more about the Department of Radiology