How low can you go? Analytical performance of five automated testosterone immunoassays.

BACKGROUND: Testosterone is commonly measured using immunoassays, yet concerns with the accuracy and quality of testing by these methods exist, particularly for low testosterone concentrations. Study objectives were to evaluate selective performance characteristics, including functional sensitivity (FS), of 5 automated immunoassays for total testosterone. METHODS: FS, imprecision, assay interference, limit of blank, linearity, and accuracy were assessed using the Abbott ARCHITECT i2000SR, SIEMENS ADVIA Centaur and IMMULITE 2000, Beckman Coulter DxI 800, and Roche MODULAR E170. Comparisons to an in-house liquid chromatography-tandem mass spectrometry (LC-MS/MS) method were performed using patient samples from men, women, boys, and girls. RESULTS: FS at 20% coefficient of variation (CV) for the ARCHITECT, Centaur, DxI, E170 and IMMULITE assays were 0.14, 1.23, 0.36, 0.77, 3.49 nmol/L, respectively. Total CVs for the 5-day imprecision study were ≤ 9.0% for all methods. All assays met manufacturer's claims for hemolysis, icterus, and lipemia interference and limit of blank. Dilution linearity studies had deviations from the target recoveries ranging from 3.4% (ARCHITECT) to 14.3% (DxI). Using National Institute of Standards and Technology Standard Reference Material 971, recoveries ranged from 79.2-149.2% (DxI, male and female, respectively). When compared to LC-MS/MS, more immunoassays under-recovered in men and women and over-recovered in boys and girls. Slopes ranged from 0.71 (IMMULITE, women) to 1.35 (DxI, boys). The combined average for percent bias was higher in boys (28.0%) than men (11.6%), women (22.8%), and girls (25.7%). CONCLUSIONS: Challenges with accurately measuring testosterone appear to remain for some immunoassays, but not all. While most immunoassays remain optimized for concentrations observed in healthy men, some showed acceptable performance when challenged at lower concentrations.