Interlaboratory/intermethod differences in functional sensitivity of immunometric assays of thyrotropin (TSH) and impact on reliability of measurement of subnormal concentrations of TSH.

Clinically relevant interassay precision profiles for thyrotropin (thyroid-stimulating hormone; TSH) were constructed with human serum pools measured over 4-8 weeks by six immunometric assays, in at least two different reagent lots. Functional sensitivities (the concentration at which the interassay CV is < or = 20%) were determined in four to eight clinical laboratories plus the respective manufacturer's laboratory. These studies revealed that the manufacturer's stated functional sensitivity limit is rarely duplicated in clinical practice. Loss of specificity (indicated by artifactually high values) was seen with some methods when used to measure certain unrefrigerated low-TSH sera. Measurement of TSH in four human serum pools (TSH < 0.05-0.25 mIU/L) by 16 different methods (each in at least eight UK or US laboratories) showed that some methods could not reliably distinguish subnormal from normal TSH values. Better pool rankings and fewer misclassifications of low-TSH sera as "normal" were seen with use of assays capable of "third-generation" functional sensitivity (0.01-0.02 mIU/L) than with assays with "second-generation" functional sensitivity (0.1-0.2 mIU/L). Because inter- and intramethod differences in functional sensitivity negatively impact the diagnostic accuracy and cost-effectiveness of a TSH-centered thyroid-testing strategy, laboratories should independently establish an assay's functional sensitivity by a clinically relevant protocol. Moreover, manufacturers should assess functional sensitivity more realistically and improve the robustness of assays to ensure that their performance potential is consistently met in clinical practice.