Evaluation of the Siemens ADVIA Centaur high-sensitivity cardiac troponin I assay in serum

BACKGROUND: International recommendations on high-sensitivity cardiac troponin (hs-cTn) testing recommend that laboratories select only one sample type for testing. We evaluated the Siemens ADVIA Centaur hs-cTnI assay in serum and thereby provide information on imprecision, long-term storage stability, freeze-thaw stability, method comparison to other hs-cTnI assays, and clinical performance. METHODS: Patients with chest pain onset <6 h who already had Roche hs-cTnT, Beckman hs-cTnI and Abbott hs-cTnI results recorded and had non-thawed and frozen serum aliquots formed the study population (n = 134 patients with 305 serum aliquots obtained at either 0, 3 or 6 h stored below -70 °C since 2003) for measurement with the Siemens hs-cTnI assay in 2018. Receiver-operating characteristic curve analyses for myocardial infarction (MI) using the highest obtained hs-cTn concentration was performed. Additional comparison testing on serum samples stored frozen (at -70 °C for <1 month in 2018) for the Siemens and Abbott hs-cTnI assays were performed, as well as precision testing in serum pools and freeze-thaw stability testing. RESULTS: The Siemens hs-cTnI assay had an area under the curve (AUC) of 0.978 (95%CI: 0.937-0.996) for MI in the study cohort (Roche hs-cTnT AUC = 0.965 and Abbott AUC = 0.973). The Siemens hs-cTnI assay yielded higher cTnI concentrations than the other hs-cTn assays, with the same proportional bias (slope = 1.4) between the Siemens and Abbott hs-cTnI assays obtained from serum samples collected in 2003 and 2018. Over 3 months, a low serum pool of 3.5 ng/l achieved a CV of 20% (SD = 0.7, n = 42) and a high serum pool of 820 ng/l achieved a CV of 2.3% (SD = 20, n = 42). Three different serum pools recovered within 10% from baseline concentration after 5 freeze-thaw cycles for the Siemens hs-cTnI assay. CONCLUSIONS: In serum, the Siemens ADVIA Centaur hs-cTnI assay had excellent clinical performance for MI in an early chest pain onset population, acceptable precision at normal and highly elevated cTnI concentrations, long-term storage stability (15 y storage below -70 °C) and acceptable freeze-thaw stability, all of which supports serum as an acceptable sample type to use in clinical studies and in clinical practice.