Sensitive thyrotropin assays: analytic and clinical performance criteria

Abstract

Thyrotropin (TSH) measurements in serum traditionally have been used for the diagnosis and management of hypothyroidism; the availability of highly sensitive TSH assays has expanded the utility of these measurements to include detection of hyperthyroidism, monitoring of thyroxine replacement therapy, and front-line assessment of thyroid function. The analytic advancement that has expanded TSH utility is the noncompetitive "sandwich" immunoassay, which is faster and more sensitive than traditional radioimmunoassays. All such assays, however, are not equivalent; therefore, we have developed analytic and clinical performance criteria for evaluating sensitive TSH assays. These criteria are as follows: (1) the overlap between the assay detection limit and the lower limit of normal should be less than 1%, (2) basal TSH measurements should predict the TSH response to thyrotropin releasing hormone stimulation at least 95% of the time, and (3) basal TSH measurements should be 95% sensitive and 95% specific for detecting hyperthyroidism. Three allegedly sensitive TSH assays were evaluated: (1) the Hybritech two-step Tandem-R method (assay A), (2) the semiautomated Boots-Celltech Sucrosep method (assay B), and (3) a modified Ciba Corning Magic Lite method (assay C). Both assay B and assay C fulfilled out criteria for a sensitive assay; assay B had slightly better performance, and assay C was easier to perform.