How Do We Measure Hyperandrogenemia in Patients With PCOS?

The latest guidelines recommend that hyperandrogenemia should be evaluated biochemically in all women suspected of having PCOS (1). There is currently no consensus on what is the most appropriate androgen to measure or the upper cutoff consistent with PCOS, but it has been generally accepted until now that T is the most commonly ordered measurement for the investigation of female hyperandrogenemia. This concept is challenged by O’Reilly et al (2) in this issue of the JCEM. It has previously been reported that androstenedione (A) can be raised when T is normal in patients with hirsutism (3) and PCOS (4), but O’Reilly et al (2) now show that patients with high A and normal T concentrations have nearly as much risk for metabolic disease as those with a high T concentration alone. The results of this study raise concern that hyperandrogenemia may be missed if only T is measured. Importantly, the serum A and T results were measured using liquid chromatography-tandem mass spectrometry (LCMS/MS). LC-MS/MS methods for measuring steroids represent the state of the art and can reliably measure accurate concentrations of A and T in female subjects. The concentration of T in females is 15 to 20 times lower than that found in males, and this has always posed a major challenge to “direct” immunoassay methods. Most clinical chemistry laboratories perform “direct” T immunoassays (without any extraction step) routinely on automated analyzers to improve the throughput and cost of analysis. Extraction steps using organic solvents had been popular and were very effective for removing interfering substances before analysis by immunoassay methods, but they are time-consuming and expensive in staff time. The extraction steps were discarded to streamline the process, but the resultant direct methods, although convenient, could suffer significant interference from other steroids such as dehydroepiandrosterone sulfate (DHEAS). DHEAS is the most abundant steroid produced by the adrenal gland and is present in 1000-fold higher concentrations than T, and the resultant positive interference increased inaccuracy of the result (5, 6). T measurement by direct immunoassay is also limited by a lack of sensitivity and specificity, which has been well documented and has prompted a recommendation from The Endocrine Society to avoid using such assays (7). In addition, the continuing poor performance of many of these direct assays has prompted a recent editorial in the JCEM (8), stating that LC-MS/MS is a mandatory submission requirement for sex steroid results as of January 2015. A is not as widely measured in clinical chemistry laboratories as T. This is partly due to demand because A is often used as a secondary test to be ordered only when T is elevated and also because A is not routinely available on main clinical chemistry analyzers. The smaller niche immunoassay analyzers available for measuring A typically give results twice the concentration of those reported by LC-MS/MS methods (9). A consequence of reporting these higher immunoassay results is that the diagnostic cutoffs reported by O’Reilly et al (2) are not applicable, thus necessitating the need for separate assay-specific reference intervals. Whether properly constructed reference intervals applicable to specific populations exist for these assays is debatable. The argument for measuring A only on samples with high T results, usually to save money and resources when using immunoassay, is now flawed because it is clear that measuring A only when T was raised