Comparison of three serum androstenedione assays: the Siemens Immulite immunoassay, the Roche Elecsys immunoassay, and an LC-MS/MS assay

Abstract

Introduction Serum androstenedione (ASD) is a useful biomarker in the diagnostic workup of hyperandrogenism, congenital adrenal hyperplasia, premature adrenarche, and polycystic ovary syndrome (PCOS). Recently, the Elecsys ASD assay (Roche Diagnostics), a competitive electrochemiluminescence immunoassay, became available in the US. Herein, the analytical and clinical performance of the Elecsys ASD assay was tested and compared with the Immulite assay (our current assay) and an LC-MS/MS assay (the gold standard) using deidentified residual patient specimens. Method In this study, the linearity, analytical measuring range (AMR), precision, and accuracy of the Elecsys ASD assay (cobas e602 analyzer) were tested. ASD from 40 deidentified residual serum/plasma samples was measured and compared between the Elecsys assay, the Immulite assay, and an LC-MS/MS assay. The reference intervals (RIs) provided by Roche for healthy male (0.280-1.52 ng/mL), healthy female (0.490-1.31 ng/mL), postmenopausal female (0.187-1.07 ng/mL), healthy children (<0.519 ng/mL), and patients with PCOS (0.645-3.47 ng/mL) were tested with at least 20 specimens, according to CLSI C28A3. Statistical analysis was performed using EP evaluator and R program. Result and conclusion The assay had a linear response across the AMR (0.3-10.0 ng/mL). The inter- and intra-assay coefficients of variation measured at multiple concentrations were less than 4.5% and 2.0%, respectively. The Elecsys ASD assay had an excellent correlation with the LC-MS/MS assay with a mean bias of -0.0542 ng/mL (-2%). The Immulite assay had a mean bias of 1.15 ng/mL (44%) and 1.22 ng/mL (32%) compared to the LC-MS/MS and Elecsys ASD assays, respectively. The Roche recommended RIs for healthy males and postmenopausal females were successfully verified in our patient population. However, the ASD concentrations for the healthy children were outside of the suggested RI, with concentrations up to 1.41 ng/mL. Therefore, the RIs for healthy children were adopted from RIs established using the same LC-MS/MS method used for method comparison. RI verification for the healthy female group also failed since many low ASD values were observed. Instead, a RI of < 1.30 ng/mL was established using 60 deidentified residual serum/plasma specimens. Finally, a separate RI for the PCOS group was not established since it may not provide useful clinical information due to the heterogeneity of the group. Unlike some published studies, hormone therapies such as oral contraceptive pills did not cause a significant decrease in ASD in patient specimens (p=0.4967). Overall, the Elecsys ASD assay has superior comparability to the LC-MS/MS assay than the Immulite assay. We were unable to verify the applicability of the RIs recommended by Roche for healthy females and children, warranting the need to establish or transfer them. When RI verification is challenging due to limited qualified specimens, transferring from an LC-MS/MS established RI is possible as long as the methods are comparable.

Design and Validation of a Sensitive Multisteroid LC-MS/MS Assay for the Routine Clinical Use: One-Step Sample Preparation with Phospholipid Removal and Comparison to Immunoassays

“…In the light of high accuracy of the LC-MS/MS method in the validation experiments (UK NEQAS proficiency testing), it seems obvious that the found differences are mainly caused by analytical problems of the IA. Several previous studies and data of the UK NEQAS scheme are available for the Siemens assay, which all clearly substantiate our findings for the high inaccuracy in measuring ANDRO [ 56 , 59 , 60 ].…”

“…There are initiatives of IA manufacturers for further enhancements by calibration against reference methods. For example, a new assay for ANDRO was announced by Roche that demonstrated much better agreement with LC-MS/MS than previous IAs [ 59 ].…”