Comparison of methods for assessing testosterone concentrations in a population-based sample of men

Program: Abstracts - Orals, Featured Poster Presentations, and Posters
Session: SUN 524-553-Male Reproductive Endocrinology
Bench to Bedside
Sunday, June 16, 2013: 1:45 PM-3:45 PM
Expo Halls ABC (Moscone Center)

Poster Board SUN-543
Andre B. Araujo*1, Carrie G. Wager2 and Gary Allen Wittert3
1New England Research Institutes, Watertown, MA, 2New England Research Institutes, Inc., Watertown, MA, 3Royal Adelaide Hospital, Adelaide SA, Australia
Clinically, it is generally accepted that high-quality radioimmunoassays (RIA) or platform-based chemiluminesence immunoassays (CI) provide sufficient information on serum testosterone levels (T) in aging men, but such may not be the case in the research setting. We examined the relationships between 7 methods of assessing T and their ability to predict androgen-sensitive endpoints in a community-based survey of 1195 men age 35-80y from Adelaide, South Australia. T levels were assessed by CI (total), liquid chromatography-tandem mass spectrometry (LCMS, total), measured bioavailable (mBioT) by ammonium-sulfate precipitation, calculated bioavailable based on CI (cBioT), calculated free T (ftCI, ftLCMS), and a bioactive assay.1 Phenotypic endpoints included isometric grip strength (IGS), DXA-derived lean mass (LM) and fat mass (FM), hematocrit (Hct), and sexual desire. T measures were compared using Pearson correlations and Bland-Altman analysis. Multiple correlations were computed for each endpoint-T measure pair using a linear model of the unit-scaled endpoint regressed on the unit-scaled log T measure with relevant adjustments for age, demographics, body composition, smoking, physical activity, alcohol, obesity, depression (Beck), self-rated health (SF36), and medication use. Stratification was performed for age (≥50) and obesity (BMI≥30). Mean age±SD of the cohort was 55±12y, and mean T by LCMS and CI were 17.4±6.6nmol/L and 13.9±5.5nmol/L, respectively. Correlations with LCMS were as follows: CI, 0.78; mBioT, 0.57; cBioT, 0.74; bioactive, 0.59; ftCI, 0.47; ftLCMS, 0.72. Bland-Altman plots showed a negative bias which was not concentration-dependent for the log ratio between LCMS and other T measures. Considering significant relations of T with endpoints: CI was associated with FM (beta: -0.11), Hct (0.04), IGS (0.14), and sexual desire (0.29); LCMS was associated with LM (0.05), Hct (0.09), and sexual desire (0.27); ftCI was associated with FM (-0.07), IGS (0.07), and sexual desire (0.16); and ftLCMS was associated with FM (-0.03), Hct (0.03), and sexual desire (0.14). Overall, total fractions of T (vs. free) and free fractions based on CI (vs. those based on LCMS) exhibited more significant relations with phenotypic endpoints. These data suggest that T measured by standard assays provides clinically relevant information. Consideration of SHBG aids interpretation particularly in older or obese men, but calculation of free T confers no useful additional information.

(1) Need EF, O'Loughlin PD, Armstrong DT, Haren MT, Martin SA, Tilley WD; Florey Adelaide Male Aging Study, Wittert GA, Buchanan G. Serum testosterone bioassay evaluation in a large male cohort. Clin Endocrinol (Oxf). 2010 Jan;72(1):87-98.

Disclosure: ABA: Principal Investigator, Abbott Laboratories, Principal Investigator, GlaxoSmithKline, Consultant, Lilly USA, LLC. GAW: Principal Investigator, Bayer Schering Pharma, Principal Investigator, Eli Lilly & Company, Advisory Group Member, Eli Lilly & Company, Researcher, Lawley Pharmaceuticals, Researcher, Lawley Pharmaceuticals, , Bayer, Inc., , Eli Lilly & Company, , Novo Nordisk. Nothing to Disclose: CGW

*Please take note of The Endocrine Society's News Embargo Policy at

Sources of Research Support: Supported by NHMRC project grant 627227, South Australian Premiers Science Award, Florey Foundation, University of Adelaide.