Sex-Specific 99th Percentile Upper Reference Limits for High Sensitivity Cardiac Troponin Assays Derived Using a Universal Sample Bank

• Published in: Clinical chemistry (Baltimore, Md.) Vol. 66; no. 3; pp. 434 - 444
• Main Authors: Apple, Fred S, Wu, Alan H B, Sandoval, Yader, Sexter, Anne, Love, Sara A, Myers, Gary, Schulz, Karen, Duh, Show-Hong, Christenson, Robert H
• Format: Journal Article
• Language: English
• Published: England American Association for Clinical Chemistry, Inc 01.03.2020; Oxford University Press
• Subjects: African Americans; Assaying; Banks (Finance); Biomarkers; Calcium-binding protein; Data buses; Epidermal growth factor receptors; Ethylenediaminetetraacetic acid; Glycosylated hemoglobin; Heart; Hemoglobin; Laboratories; Medical screening; Men; Normality; Sensitivity; Sex; Statins; Statistical methods; Statistics; Troponin; Women; Womens health; Israel; United States > US
• ISSN: 0009-9147; 1530-8561; 1530-8561
• DOI: 10.1093/clinchem/hvz029

How to select healthy reference subjects in deriving 99th percentiles for cardiac troponin assays still needs to be clarified. To assist with global implementation of high sensitivity (hs)-cardiac troponin (cTn) I and hs-cTnT assays in clinical practice, we determined overall and sex-specific 99th percentiles in 9 hs-cTnI and 3 hs-cTnT assays using a universal sample bank (USB). The Universal Sample Bank (USB) comprised healthy subjects, 426 men and 417 women, screened using a health questionnaire. Hemoglobin A1c (>URL 6.5%), NT-proBNP (>URL 125 ng/L) and eGFR (<60 mL/min), were used as surrogate biomarker exclusion criteria along with statin use. 99th percentiles were determined by nonparametric, Harrell--Davis bootstrap, and robust methods. Subjects were ages 19 to 91 years, Caucasian 58%, African American 27%, Pacific Islander/Asian 11%, other 4%, Hispanic 8%, and non-Hispanic 92%. The overall and sex-specific 99th percentiles for all assays, before and after exclusions (n = 694), were influenced by the statistical method used, with substantial differences noted between and within both hs-cTnI and hs-cTnT assays. Men had higher 99th percentiles (ng/L) than women. The Roche cTnT and Beckman and Abbott cTnI assays (after exclusions) did not measure cTn values at ≥ the limit of detection in ≥50% women. Our findings have important clinical implications in that sex-specific 99th percentiles varied according to the statistical method and hs-cTn assay used, not all assays provided a high enough percentage of measurable concentrations in women to qualify as a hs-assay, and the surrogate exclusion criteria used to define normality tended to lower the 99th percentiles.