Abstract

SCAGLIA, HE et al. A Comparison of Serum Testosterone (T) Measurements by Various Methods, Employing Current Validated Techniques or Otherwise, and Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS): A Multicentric Study. Rev. argent. endocrinol. metab. [online]. 2015, vol.52, n.3, pp.137-152. ISSN 1851-3034.

Objective: To compare normal and hirsute women Testosterone (T) measurements performed at different laboratories by the same or different methods, and the gold standard method LC-MS/MS (Quest Diagnostics, USA). Design: Prospective study. Setting: Hormone Determination Laboratory, Hospital Italiano, La Plata, and each participating laboratory's private practice. Patient(s): Blood samples were obtained from 23 individuals sorted into two groups, namely, normal women, n: 11(NW) and hirsute women, n: 12 (HW). Interventions(s): None. Main Outcome Measure(s): To evaluate whether serum T measurements obtained from each serum by the methods currently employed in our country, some of whose kits exhibit changes in previous presentations, some LC-MS/MS-validated and other non-validated ones are significantly different from those obtained by LC-MS/MS. Result(s): None of the 11 NW showed high T values by LC-MS/MS. Two out of the 12 hirsute patients showed normal T values (LC-MS/MS). Methods and number of participating labs -shown between brackets were: in NW, 1^st generation Architect (1), 2^nd generation Architect (1); Immulite (1) Cobas (4); Access (1); Centaur (2); Immunotech-RIA (1); and, in HW, 2^nd generation Architect (3); Immulite (3); Cobas (4); Access (1); Centaur (2); Immunotech-RIA (1). No false positives resulted from the assays performed. No lab yielded false positive results in the NW group. No false positives were reported from the 10 hirsute women with increased T values by LC-MS/MS. False positives, though, resulted from two female hirsute patients with normal T values studied by four of the methods. Statistically, the serum T measurements obtained were significantly different by Centaur in NW and, in HW, by Immulite and Centaur as compared to LC-MS/MS. In the Bland-Altman plot, Centaur and Cobas showed over 5 % of measurements outside the limits of agreement in the HW group. Assessment by p-Spearman resulted in divergences with LC-MS/MS for all methods in NW, whereas in the HW group there were none. When estimating sampling bias for each laboratory taking LC-MS/MS as the reference method and adopting a ± 6.4 % mean bias acceptability criterion for each method compared to LC-MS/MS, two of the techniques reviewed, 2^nd generation Architect and Cobas, met the validation requirement satisfactorily. However, one lab out of three using 2^nd generation Architect failed to meet the validation requirement, while two out of four labs using Cobas also failed to meet the requirement. This demonstrates the great variability among methods, even when labs are employing the same technique. Conclusion: From the clinical point of view, the methods currently used in our local environment yielded no false positives or false negatives and therefore did not misdiagnose hyperandrogenism. Still, Immulite, Centaur, RIA and Access did present false positives in two of the T-normal hirsute women. The relation of serum T measurements obtained by each method to measurements obtained by LC-MS/MS reveals that the dispersion of the results was larger with values under 0.3 ng/ml, quite close to the detection limit of the various techniques.

Keywords : Testosterone; Testosterone assay; LC-MS/MS.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )