Fast 0/1-h algorithm for detection of NSTEMI: are current high-sensitivity cardiac troponin assays fit for purpose? An EQA-based evaluation

• Published in: Clinical chemistry and laboratory medicine Vol. 57; no. 12; p. 1999
• Main Authors: van der Hagen, Eline A E, Cobbaert, Christa M, Meijer, Ron, Thelen, Marc H M
• Format: Journal Article
• Language: English
• Published: Germany 26-11-2019
• Subjects: Algorithms; Bias; Biological Assay - methods; Blood Chemical Analysis - methods; Female; Humans; Male; Myocardial Infarction - diagnosis; Non-ST Elevated Myocardial Infarction - diagnosis; Non-ST Elevated Myocardial Infarction - metabolism; Reference Values; Reproducibility of Results; Sensitivity and Specificity; Time Factors; Troponin I - analysis; Troponin T - analysis; diagnosis; high-sensitivity methods; analytical specifications; cardiac troponin; EQAS; myocardial infarction
• ISSN: 1437-4331
• DOI: 10.1515/cclm-2019-0253

Background High-sensitivity cardiac troponin T/I (hs-cTnT/I) assays have improved analytical sensitivity for the detection of myocardial infarction (MI). To gain clinical specificity and sensitivity, interpretation of changes in cTn concentrations over time is crucial. The 2015 ESC NSTEMI guideline defines absolute delta values as additional rule-in and rule-out criteria for MI. A critical assumption for application of this rule is that total analytical imprecision within the delta period, including inter-instrument bias, is comparable to analytical imprecision in the validation studies. Methods Data from the Dutch External Quality Assessment Scheme (EQAS) were used to calculate inter-instrument bias and estimate imprecision for the measuring range where the proposed delta values are relevant: for Roche Elecsys hs-cTnT, 5-52 and 5-12 ng/L; for Abbott Architect hs-cTnI, 2-52 and 2-5 ng/L for rule-in and rule-out, respectively. Results For Elecsys, the median inter-instrument bias is 0.3 ng/L (n = 33 laboratories), resulting in reference change values (RCVs) of 3.0 and 1.7 ng/L, respectively, for rule-in and rule-out with imprecision as claimed by the manufacturer. With RCVs smaller than the guideline's delta thresholds, 100% of the laboratories have adequate specifications. RCVs for rule-in/rule-out increased to 4.6 ng/L/2.5 ng/L, respectively, with individual imprecisions as estimated from EQA data, resulting in 64% and 82% of laboratories with adequate specifications. For Architect, 40% of instruments (n = 10) might falsely qualify the result as clinically relevant; hence, inter-instrument bias could not be determined. Conclusions We advise laboratories that use the fast 0/1-h algorithm to introduce stringent internal quality procedures at the relevant/low concentration level, especially when multiple analyzers are randomly used.