Cardiac troponins T and I: reproducible discrepancies in the clinical setting.

The universal definition of myo-cardial infarction includes the increase and/or decrease in cardiac biomarkers, with at least 1 value greater than the 99th-percentile upper reference limit, concurrently with evidence of myocardial isch-emia and corresponding clinical symptoms, electrocardiographic changes, or imaging evidence (1). Among cardiac biomarkers, cardiac troponins, including cardiac troponin I (cTnI) 1 and cardiac tro-ponin T (cTnT), have become the gold standard. Several noncardiac etiologies for increases in cardiac troponins have been described, however (2), and discordant values for cTnT and cTnI have been encountered in the clinical setting. The aim of our study was to systematically compare a cTnT assay and a cTnI assay with regard to dis-crepant values. This study was conducted in accord with the World Medical Association Declaration of Helsinki (3). We used the Architect cTnI STAT Troponin-I assay (Abbott Diagnostics) and the new high sensitive troponin T (hsTnT) assay (Troponin T hs) on the Modular Analytics system (Roche Diagnos-tics) to make 9004 simultaneous cTnI and cTnT measurements on 3995 patients and then analyzed the data. Because we were concerned about major discrepancies, we used 0.032 ␮g/L as the cutoff value for both the cTnI and hsTnT measurements. After applying this cutoff to our samples, we found that 1037 samples from 664 patients had inconsistent cTnI and hsTnT results. The majority of the dis-crepant results were explained by minor differences near the cutoff or by differences in cardiac tropo-nin kinetics. To exclude minor inconsistencies , we filtered our database to retain only entries in which a result for one of the cardiac troponin assays was below the 0.032-␮g/L cutoff value and the result for the other assay was Ͼ0.096 ␮g/L (i.e., 3 times the cutoff value). This filtering produced 128 discrepant pairs of results, from 83 cases. To minimize the possibility of measurement discordance attributable to dynamic differences or sampling errors, we examined only cases of discrepant results that were confirmed by at least one additional measurement from a separate blood sample collected during the same hospital stay. This second filtering step produced 18 cases, which we evaluated in detail. Nine of the 18 cases had cTnI concentrations above and hsTnT concentrations below our cutoffs (Table 1, case nos. 1–9). The primary clinical diagnoses in this group were acute coronary syndrome (ACS) (n ϭ 2), coronary artery disease without ACS (n ϭ 4), hypertension (n ϭ 5), atrial fibril-lation (n ϭ 1), respiratory diseases (n …