Comparison of 13 Commercially Available Cardiac Troponin Assays in a Multicenter North American Study. (1st March 2017)

• Record Type: Journal Article
• Authors: Christenson, Robert H; Jacobs, Ellis; Uettwiller-Geiger, Denise; Estey, Mathew P; Lewandrowski, Kent; Koshy, Thomas I; Kupfer, Kenneth; Li, Yin; Wesenberg, James C
• Is Part Of: Journal of applied laboratory medicine. Volume 1:Number 5(2017)
• Page Start: 544
• Page End: 561
• Publication Date: 2017-03-01
• Subjects: Clinical chemistry -- Periodicals; Diagnosis, Laboratory -- Periodicals; 616.0756
• DOI: 10.1373/jalm.2016.022640
• Languages: English
• ISSNs: 2576-9456

Abstract

Abstract: Background: We examined the concordance of 13 commercial cardiac troponin (cTn) assays [point-of-care, high-sensitivity (hs), and conventional] using samples distributed across a continuum of results. Methods: cTnI (11 assays) and cTnT (2 assays) were measured in 191 samples from 128 volunteers. cTn assays included Abbott (iSTAT, STAT, and hs), Alere (Cardio 3), Beckman (AccuTnI+3), Pathfast (cTnI-II), Ortho (Vitros), Siemens (LOCI, cTnI-Ultra, Xpand, Stratus CS), and Roche [4th Generation (Gen), hs]. Manufacturer-derived 99th percentile cutoffs were used to classify results as positive or negative. Alternative 99th percentile cutoffs were tested for some assays. Correlation was assessed using Passing–Bablok linear regression, bias was examined using Bland–Altman difference plots, and concordance/discordance of each method comparison was determined using the McNemar method. Results: Regression slopes ranged from 0.63 to 1.87, y -intercepts from 0.00 to 0.03 ng/mL, and r values from 0.93 to 0.99. The cTnT methods had a slope of 0.93, y -intercept of 0.02 ng/mL, and r value of 0.99. For the cTnI assays, positive, negative, and overall concordance was 76.2%–100%, 66.0%–100%, and 82.9%–98.4%, respectively. Overall concordance between the 4th Gen cTnT and hsTnT assays was 88.9%. A total of 30 of the 78 method comparisons showed significant differences in classification of samples ( P <0.001); the iSTAT showed 10, hsTnT showed 9, AccuTnI+3 showed 5, Xpand showed 5, and Stratus CS showed 1. Using alternative 99th percentile cutoffs to those listed by manufacturers lowered the method discordance by 6-fold, from 30 to 5 (all involved iSTAT). Conclusions: These data provide insight into characteristics of cTn methods and will assist the healthcare community in setting expectations for relationships among commercial cTn assays.