The Testosterone Assay: Fact or Fantasy

Abstract

The clinical endocrinologist requires accurate, reproducible, and readily available hormone assays for diagnosis and management of endocrine disorders. Probably nowhere in clinical endocrinology is the inadequacy of commercially available endocrine assays more apparent than in the measurement of serum testosterone when applied to the diagnosis of hyperandrogenism in women. The assay of testosterone is technically challenging. The most exacting method, and therefore the standard for measuring testosterone, involves gas-liquid chromatography or high-performance liquid chromatography (HPLC) followed by mass spectrophotometry. This method, however, is not cost-effective for the commercial laboratory. Immunoassays provide efficient and cost-effective systems with adequate specificity and sensitivity to measure testosterone accurately. Because of the small molecular size of steroids, antibodies induced against these compounds tend to be less specific than those induced with larger protein molecules. This factor leads to the necessity of extracting and chromatographing the sample before antibody exposure to distinguish among steroids of similar structure (1). In order to use radioimmunoassay (RIA) for accurate measurement of testosterone, chromatographic purification of the extracted sample is needed for most assays. Without adequate sample preparation, many RIA systems will substantially overestimate the true testosterone level—especially those systems that use antibodycoated tubes, which is a common method in commercial laboratories. Accurate measurement of testosterone, however, is not easy. A recent report from Falk et al. (2) evaluated the coefficients of variation for replicate quality control samples in a single academic laboratory. Intraclass correlation of the testosterone assay was only fair (approximately 50%). The diagnosis of hyperandrogenism is based on increased androgen levels, and the patient’s serum level is compared with a reference range of values. A review of the scientific literature on hyperandrogenism showed that the reference range of androgen values is established by each research laboratory, and the authors must remind the reader that the application of the published data to the clinician’s practice can be made only when the reader knows the methods and reference ranges for the commercial laboratory used by the clinician. Obtaining this information from the commercial laboratory is an impossible task, as attested to by Steinberger and coworkers in this issue of Endocrine Practice(pages ?? to ??). Although they do not state the obvious, clearly the supposed reference range provided by the commercial laboratories is simply a restating of the numbers provided by the RIA kit manufacturer. Neither the manufacturer nor the laboratory has made any attempt to perform the due diligence necessary to verify that the method is clinically relevant or accurate. Steinberger and coworkers’ statistical analysis of the scientific literature leaves no doubt that testosterone values in normal women and those with hyperandrogenism are clearly separated when data from clinical research lab