Temperature effects on free-thyroxine measurements: analytical and clinical consequences.

We compared the analytical and clinical performance of two free-thyroxine (FT4) assays--a solid-phase radioimmunoassay, Spectria, and a time-resolved fluoroimmunoassay, Delfia, both of them two-step methods--with the performance of a direct radioimmunoassay, Nichols, to measure FT4 concentration in equilibrium dialysate of undiluted serum. The three assays showed comparable analytical performance. We tested clinical utility in sera from 135 healthy subjects with and without thyroxine-binding abnormalities and in 61 patients with and without thyroidal illnesses. We found significant differences for FT4 measured by different assays in sera from the same euthyroid patients. To explain the differences, we studied the influence of temperature on performance and calibration. Most important was the neglected fact that the association constant for the binding of thyroxine to thyroxine-binding globulin decreases when the temperature rises from 20 to 37 degrees C, causing a doubling of FT4. The two-step assays, if performed at room temperature without a well-defined calibration, can give misleading FT4 concentrations. This is the case when sera from patients with thyroxine-binding abnormalities are measured against kit standards, made up in normal human sera. If an assay is to reflect the in vivo FT4 concentration at body temperature in all types of samples, it should be performed at body temperature. For practical reasons 37 degrees C is recommended, and reference values should be defined at 37 degrees C. The same might be valid for other free-hormone assays.