RESUMO
BACKGROUND: Accurate serum total thyroxine (TT4) measurement is important for thyroid disorder diagnosis and management. We compared the performance of six automated immunoassays with that of isotope-diluted liquid chromatography-tandem mass spectrometry (ID-LC-MS/MS) as the reference method. We also evaluated the correlation of thyroid stimulating hormone (TSH) with TT4 measured by ID-LC-MS/MS and immunoassays. METHODS: Serum was collected from 156 patients between October 2015 and January 2016. TT4 was measured by immunoassays from Abbott (Architect), Siemens (ADVIA Centaur XP), Roche (E601), Beckman-Coulter (Dxi800), Autobio (Autolumo A2000), and Mindray (CL-1000i), and by ID-LC-MS/MS. Results were analyzed using Passing-Bablok regression and Bland-Altman plots. Minimum requirements based on biological variation were as follows: a mean bias of ≤4.5% and total imprecision (CV) of ≤3.7%. RESULTS: All immunoassays showed a correlation >0.945 with ID-LC-MS/MS; however, the slope of the Passing-Bablok regression line varied from 0.886 (Mindray) to 1.23 (Siemens) and the intercept from −12.8 (Siemens) to 4.61 (Mindray). Only Autobio, Beckman-Coulter, and Roche included the value of one in the 95% confidence interval for slope. The mean bias ranged from −10.8% (Abbott) to 9.0% (Siemens), with the lowest value noted for Roche (3.5%) and the highest for Abbott (−10.8%). Only Abbott and Roche showed within-run and total CV ≤3.7%. CONCLUSIONS: Though all immunoassays correlated strongly with ID-LC-MS/MS, most did not meet the minimum clinical requirement. Laboratories and immunoassay manufacturers must be aware of these limitations.