Objective: We measured the changes in testosterone fractions in serum of normal men over a 24-hour period, and determined whether they could be simulated on the basis of current understanding of the interactions between steroids and binding proteins in the blood.
Design: Starting from between 0830 and 0930 h, blood samples were taken every 45 minutes for 25.5 hours.
Patients: Five healthy males aged 26-45 years. All participants worked on a hospital campus and while being sampled carried out their normal activities during waking hours.
Measurements: The concentrations of testosterone (RIA) and albumin, and the percentage non-sex hormone binding globulin-bound testosterone (ammonium sulphate precipitation) and percentage free testosterone (rate dialysis), were measured on each sample. Cortisol (RIA) and sex hormone-binding globulin (SHBG) (IRMA) concentrations were measured on every second sample, and that of corticosteroid-binding globulin on two samples from each series.
Results: In all participants the levels of free and non-SHBG-bound testosterone in early morning samples (near 0530 h) were significantly different from those taken before midnight (P < 0.0005). Significant circadian rhythms (P < 0.05) in the concentration of testosterone and in the level of the free fraction were detected in all participants, and in four of the five participants for the non-SHBG-bound fraction. The amplitude of the free testosterone rhythm (34 +/- 2% of basal) was greater than that for testosterone itself (24 +/- 3% of basal). The 24-hour rhythm of the non-SHBG-bound fraction was similar to the total and free fractions except for the period 0330-0900 h when the level of this fraction declined by 15-45% over 1.5-3 hours. This decline was coincident with the initial rise in the concentration of cortisol. A decline of 10.5 +/- SEM 1.0% in the concentration of albumin, and 12.0 +/- 1.1% in that of SHBG occurred when the mean ambulant and supine levels were compared; analysis indicated significant circadian rhythms in the concentrations of these proteins. Simulation was used to investigate possible causes for the circadian rhythms in free and non-SHBG-bound testosterone. Simulation results matched the measured data well in qualitative terms, but quantitatively there were differences.
Conclusions: Increasing saturation of the binding proteins following rises in testosterone production, and the small but significant changes in protein concentration, probably related to postural changes, were implicated as the major factors in the rhythm amplitude. However, the early morning decline in the non-SHBG-bound fraction was not explained by these factors. The rise in cortisol concentration at this time is a probable cause. Alternatively, simulation suggests that a substance appearing in the early morning and competing with testosterone for albumin binding sites may be responsible.