Calculated free testosterone in men: comparison of four equations and with free androgen index

BACKGROUND: Serum testosterone remains the most important investigation in the diagnosis of androgen deficiency in men. Most of the circulating testosterone is bound to albumin and sex hormone-binding globulin (SHBG), whereas free testosterone accounts for approximately 2% of total testosterone. Because direct measurement of free testosterone is impractical in routine practice, several equations are used to provide clinically useful estimates of free testosterone concentration. This study aimed to (1) obtain locally derived reference limits for total testosterone and calculated free testosterone (CFT) concentrations, and (2) critically evaluate the equations commonly used to estimate free testosterone. METHODS: Serum total testosterone, SHBG and albumin were assayed in morning blood samples obtained from 126 healthy men (aged 20-45 years) known to have normal semen analysis. CFT concentrations calculated using four published methods (i.e. the Sodergard, Nanjee-Wheeler, Vermeulen and Ly-Handelsman equations) were compared with one another and the free androgen index. RESULTS: Reference intervals for total testosterone and CFT by the Vermeulen equation were 9.4-31.0 nmol/L and 0.245-0.785 nmol/L (2.5-97.5 percentile), respectively. CFT values varied considerably with the four equations examined. Mean biases ranged from 5.8 to 56.0%; the Nanjee-Wheeler and Ly-Handelsman equations yielded positive and negative biases, respectively, against the other equations. Free androgen index was shown to correlate poorly with CFT (r2=0.21-0.46) and over-estimate the CFT at low SHBG concentrations. CONCLUSIONS: We have used various equations to derive reference ranges for CFT in healthy men aged 20-45 years. We suggest that CFT be incorporated into the investigation regimen for suspected hypogonadism when total testosterone results are equivocal.