Objective: To compare results from testosterone radioimmunoassay kits commonly used by commercial laboratories as well as their reference ranges and to analyze the scientific literature for ranges of serum testosterone levels in normal women and those with hyperandrogenism.
Methods: We reviewed quality assurance reports of various testosterone ligand challenges from four groups of laboratories and summarized testosterone data from 17 published reports about normal women and 14 studies of hyperandrogenic women.
Results: A significant variability was demonstrated between the radioimmunoassay kits at all concentrations (for example, a sample with a mean testosterone level of 96.1 ng/dL was reported by some laboratories as containing 71.8 ng/dL and by others as 123.4 ng/dL). All laboratories provide essentially the same "reference range" (approximately 10 to 90 ng/dL) but do not report how the range was established. The scientific literature clearly shows a significant separation in serum testosterone levels between normal (that is, not hyperandrogenic) and hyperandrogenic women. Most hyperandrogenic women had testosterone levels >50 ng/dL, whereas most normal control subjects had levels <40 ng/dL. Thus, most of these women with hyperandrogenism would have been considered to have normal testosterone levels if the reference ranges of commercial laboratories were used.
Conclusion: These data illustrate the difficulty that physicians face when they are required to use different commercial laboratories to measure serum testosterone levels. We propose that (1) reference ranges be established on a clinically defined population for each hormone and method used, (2) laboratory reports include information about method and reference range population, and (3) physicians be allowed to choose which laboratories are used for their patients' hormone determinations, for consistency of results.