Clinical evaluation of nine free thyroxine assays: Persistent problems in particular populations

Over the past decade, numerous papers have addressed the various methodological problems encountered with free thyroxine (FT) assays. We evaluated the clinical performance of nine FT assays in five centres, using a panel of 310 sera: 156 from euthyroid controls; 27 from hyperthyroid patients; 34 from untreated hypothyroidism; 22 from patients with renal failure; 30 from women in the last trimester of pregnancy; 23 from patients on thyroid substitutive therapy; and 18 from patients treated with amiodarone. Only three methods showed a Gaussian distribution of FT concentrations. Reference ranges were calculated using the 2.5th and 97.5th percentiles. A significant difference was observed between FT values in men and women. The areas under the receiver operating characteristic (ROC) curves ranged from 0.996 to 1 for hyperthyroidism and from 0.973 to 1 for hypothyroidism. In sera from patients with renal failure and from pregnant women, method-dependent biases were observed and confirmed with dilution experiments. In conclusion, current FT assays show good performance regarding the diagnosis of overt dysthyroidism. Nevertheless, FT measurements are still vulnerable to method-dependent artefacts in particular populations such as patients with renal failure and pregnant women.