Unstable correspondence between salivary testosterone measured with enzyme immunoassays and tandem mass spectrometry

Although some studies reveal that saliva handling and storage practices may influence salivary testosterone concentrations measured with immunoassays, the effect of these method factors on the validity of testosterone immunoassays remains unknown. The validity of immunoassays can be assessed by comparing hormone concentrations measured with immunoassays to a standard reference method: liquid chromatography tandem mass spectrometry (MS). We previously reported the correspondence between salivary testosterone measured with enzyme immunoassays (EIAs) and with MS when there was less variance in (or more control over) method factors related to saliva handling and storage across measurement methods (Welker et al., 2016). In the present study, we expanded the original dataset and compared the correspondence between Salimetrics EIAs and MS when there was greater variance in (or less control over) method factors across EIAs and MS (high method variance), to when there was less variance in these factors (low method variance). If variance in these method factors impacts the validity of testosterone measurement, then the EIA-MS correspondence should be stronger when method variance is low compared to when it is high. Our results contradicted this hypothesis: Salimetrics EIA-MS correspondence was stronger when variance in method factors was high compared to when it was low. The composite average correlation across both method variance comparisons provides an updated estimate of Salimetrics EIA-MS correspondence, but the instability in this correspondence may pose challenges to the reproducibility of psychoneuroendocrinology research. We discuss possible explanations for the surprising pattern of results and provide recommendations for future research.