Not all testosterone tests are equal when PCOS is on the table

Polycystic ovary syndrome is diagnosed partly on biochemical hyperandrogenism, meaning raised androgen hormones measured in the blood. The difficulty is that several different androgens can be measured, including total testosterone, calculated free testosterone, the free androgen index, androstenedione and DHEAS, and it has never been settled which gives the most accurate answer. That uncertainty matters because a diagnosis, and everything that follows from it, can hinge on the result. This review was run specifically to inform the 2023 International PCOS Evidence-based Guidelines, asking which androgen measures and which laboratory methods most accurately identify biochemical hyperandrogenism in women being assessed for PCOS.

The team updated the evidence behind the 2018 international guidelines with a systematic search covering 3 July 2017 to 23 June 2023 across five databases. Studies evaluating androgens in the diagnosis of biochemical hyperandrogenism were assessed for risk of bias using a standard diagnostic accuracy tool. Of 23 studies reviewed, 18 supplied data for the pooled diagnostic accuracy analysis, covering 2857 participants, of whom 1650 had PCOS and 1207 were controls. For each androgen measure the reviewers calculated pooled sensitivity and specificity and built summary accuracy curves, and they compared laboratory methods, including liquid chromatography with tandem mass spectrometry against the direct immunoassay method.

Total testosterone showed a pooled sensitivity of 0.74 and specificity of 0.86, while calculated free testosterone had the highest sensitivity at 0.89 with specificity of 0.83. The free androgen index performed similarly to total testosterone. Androstenedione and DHEAS were weaker, with poorer specificity of 0.71 and 0.67 respectively, making them more prone to flagging women who do not actually have androgen excess. Laboratory method mattered too: LC-MS/MS showed superior sensitivity for several measures and superior specificity for total testosterone, calculated free testosterone and the free androgen index compared with direct immunoassay.

These results fed directly into the 2023 International PCOS Guideline recommendation to use total testosterone and free testosterone as first-line laboratory tests for biochemical hyperandrogenism, with androstenedione and DHEAS as possible additions when the first two are not elevated, bearing in mind their poorer specificity. The review also underpins the advice that laboratories should use LC-MS/MS given its higher accuracy. The clearest open question is normative cut-off values: the authors call for future work in large, unselected and ethnically diverse cohorts of women, since accuracy figures only translate into good diagnosis when thresholds fit the population being tested.

For anyone being assessed for possible PCOS, this is useful background rather than a to-do list. It explains why a clinician is likely to start with total and free testosterone rather than other androgen tests, and why two labs can give different answers on the same blood sample: the measurement technology itself differs in accuracy. It is also a reason not to read too much into a single borderline androgen result, particularly one from an older immunoassay method. The bigger message is that PCOS diagnosis rests on carefully chosen tests, and the evidence behind those choices has now been systematically pooled.