MSACL 2019 EU Abstract

Accuracy and Variability of Serum Bioavailable Testosterone Testing Methods

Yusheng Zhu, Christopher Hamilton, Kimberly Robyak
Penn State University Hershey Medical Center

Warning: Undefined variable $headshot in /var/www/html/view_abstract/view_abstract_in_program.php on line 704

Yusheng Zhu (Presenter) Penn State University >> POSTER (PDF)

Presenter Bio: Yusheng Zhu, PhD, DABCC, FAACC
Dr. Zhu is a Professor of Pathology and Pharmacology and Medical Director of Clinical Chemistry and Automated Testing Laboratory, Co-Director of Pathology Core Reference Laboratory, Director of Postdoctoral Clinical Chemistry Training Program, and Chair of Laboratory Utilization Committee at the Pennsylvania State University Hershey Medical Center. He is board certified by the American Board of Clinical Chemistry (ABCC) in Clinical Chemistry, Toxicological Chemistry, and Molecular Diagnostics and a Fellow of the AACC Academy (formerly NACB) since 2007. Currently, he is the Vice President of ABCC, Member of AACC Academy Council, and Chair-Elect of Mass Spectrometry & Separation Science Division. Yusheng is on the Editorial Boards of three medical journals and an invited reviewer of seven international medical journals. He also served as the President of the North American Chinese Clinical Chemists Association (2012), President of the Commission on Accreditation in Clinical Chemistry (ComACC), Chair of American Association for Clinical Chemistry (AACC) Southeast Section in 2015, Chair of Clinical Translational Science Division (2015-2016), and Secretary of Proteomics & Metabolomics Division (2013-2015) of AACC. He has given many presentations at national and international conferences. He is interested in clinical and translational research in clinical chemistry, toxicology, hemoglobinopathy, pharmacogenetics, proteomics, and clinical application of mass spectrometry. He has over 100 publications including peer-reviewed papers, editorials, book chapters, and abstracts. He has received 35 awards from AACC, NACB, and other organizations.

Relevant Financial Disclosures (within past 24 months) No relevant financial relationship(s) to disclose. Conflict mitigation NOT REQUIRED.

Introduction: Bioavailable testosterone (BioT) includes free and albumin-bound testosterone. Total testosterone (TT) can be measured by immunoassays and LC-MS/MS, a preferred method for women and children. BioT can be calculated using various algorithms based on TT, SHBG, and albumin concentrations or measured after removal of SHBG-bound testosterone by NH4SO4 precipitation. We compared BioT results generated by different in-house approaches with a reference lab method.
Methods: TT in 49 sera was measured by in-house LC-MS/MS, electrochemiluminescence immunoassay (ECLIA), and radioimmunoassay (RIA), respectively. SHBG was also measured by ECLIA. Fixed concentration of 43 g/L was used for albumin. BioT was calculated using four algorithms (Vermeulen, Sodergard, Emadi-Konjin, and Morris) based on TT measured by LC-MS/MS and ECLIA, SHBG, and albumin concentrations. For testosterone measured by RIA, NH4SO4 precipitation was performed. BioT values were compared with the reference lab results calculated by a proprietary algorithm using TT measured by LC-MS/MS. Deming regression was performed for method comparison and Microsoft Excel was employed for calculating means and standard deviations (SD) of percent bias. The allowable total error was ±25%.
Results: Method comparison statistics were generated for in-house LC-MS/MS (YL) and ECLIA (YE) calculated with 4 algorithms as well as NH4SO4 precipitation (YP) method compared to the reference lab (X). For LC-MS/MS, Vermeulen (YL1): YL1=0.835X–3.46 (R=0.997), mean bias=-21.1%±9.1% (mean±SD) ; Sodergard (YL2): YL2=1.078X-6.0 (R=0.996), mean bias=13.0%±14.4%; Emadi-Konjin (YL3): YL3=0.430 X-6.29 (R=0.973), mean bias=-68.5% ± 8.0%; Morris (YL4): YL4=0.540 X-4.02 (R=0.969), mean bias=-23.3%±28.8%. For ECLIA, Vermeulen (YE1): YE 1=0.798X–1.58 (R=0.9936), mean bias=-15.8% ±31.5%; Sodergard (YE2): YE2=1.034 X-5.87 (R = 0.9921), mean bias=19.4%±46.4%; Emadi-Konjin (YE 3): YE3=0.405X–3.65 (R=0.972), mean bias=-66.2%±13.0%; Morris (YE4): YE4=0.518 X–7.59 (R=0.9655), mean bias=-23.6%±34.6%. NH4SO4 precipitation (YP): YP=0.883X-10.93 (R=0.988), mean bias=-28.9% ±12.6%.
Conclusion: Compared to the reference lab BioT and based on the in-house LC-MS/MS TT, Vermeulen, Sodergard, and Morris algorithms generate values with acceptable average biases, and the Vermeulen and Sodergard algorithms have correlation coefficients >0.99. Additionally, the least variable methods are Vermeulen and Emadi-Konjin algorithms with a percent bias SD of 9.1% and 8.0%, respectively. Similarly, based on the in-house ECLIA, Vermeulen, Sodergard, and Morris algorithms also produce values with acceptable average biases, and the Vermeulen and Sodergard algorithms show correlation coefficients >0.99. However, in comparison with LC-MS/MS, BioT results calculated using ECLIA TT values demonstrate larger variability in biases compared to the reference lab BioT for all 4 algorithms. Therefore, we select the Vermeulen algorithm and LC-MS/MS TT method for calculating BioT.