Objective: To investigate performance of sex hormone binding globulin (SHBG) assays of different manufacturers and the effects of assay choice on calculated free testosterone (cFT).
Methods: Anonymized serum samples of 113 men and 106 women were randomly selected from excess material from the routine clinical lab at the University Hospital of South Manchester. SHBG levels were measured using three different immunoassays (IA): Roche (Cobas e801), Abbott (Alinity) and Siemens (Immulite). Total testosterone levels were determined at Ghent University Hospital using liquid chromatography followed by tandem mass-spectrometry (LC-MS/MS). Free testosterone (FT) levels were calculated using the Vermeulen formula for every SHBG IA. Free testosterone levels were also measured (mFT) using equilibrium dialysis followed by LC-MS/MS (ED LC-MS/MS). Results from the different assays were compared using Friedman test followed by Wilcoxon signed-rank tests with Benjamini-Hochberg correction. Method comparison for SHBG and FT was performed using Bland-Altman and Passing-Bablok analysis. The mean SHBG result and mFT were used as references for SHBG and FT respectively. FT analysis was performed separately for men and women.
Results: Mean SHBG values for all assays were 47.06 (+22.41), 45.86 (+22.22) and 44.57 (+21.04) nmoI/l for Abbott, Roche, and Siemens respectively. Mean FT values for men were 7.23 (+2.66), 7.41 (+2.70), 7.51 (+2.73) and 5.23 (12.21) ng/dl for Abbott, Roche, Siemens, and ED LC-MS/MS respectively. Mean FT values for women were 0.30 (+0.15), 0.31 (+0.15), 0.31 (10.15) and 0.22 (+0.12) ng/dl. SHBG results differed significantly between the different lAs (P< 0.001). Passing-Bablok analysis shows that the Abbott IA had a proportional difference of +3.3% in comparison to the average of the lAs while the Siemens IA had a proportional difference of -2.3%. The Roche IA did not show a significant proportional difference compared to the average. No significant constant differences were observed. cFT results similarly differed between the different lAs and compared to mFT (P< 0.001) in both men and women. In both sexes a significant proportional and constant difference was observed between mFT and all calculated equivalents. In men, the proportional difference of cFT compared to mFT ranged from +22.3 to +25.2% with a constant difference of +0.82 to +0.9 ng/dl. In women, the proportional and constant differences ranged from +24.6 to +27.9% and +0.026 to +0.029 respectively.
Conclusions: The SHBG IAs showed good agreement with only minor differences between measurements. Subsequently, the effects on FT calculations were limited with minor variation in differences between cFT and mFT. While these differences were statistically significant, they are not expected to affect clinical decision making due to their limited effect.