RIAs are based on competitive binding of testosterone to a testosterone-specific antibody. The patient’s serum is mixed with a set amount of radioactively labeled testosterone tracer and a fixed amount of antibody against testosterone. The amount of tracer displaced by the patient’s testosterone is evaluated by measuring the radioactivity of the sample, and the patient’s testosterone concentration is calculated. This method is illustrated in Fig. 2.3.

Traditional RIAs involve a testosterone extraction step, which increases their sensitivity and accuracy. These assays exhibit a variability of less than 10% in an experienced laboratory, with a lower limit of detection less than 10 pg [25]. Unfortunately, they require a technically experienced staff, are time-intensive, and generate large amounts of radioactive waste. Commercially available RIA kits are an easier alternative. However, they use a set amount of tracer and antibody, with the result that their ability to measure very high or very low levels of testosterone is somewhat poor. The detection of low levels of testosterone can be improved by increasing the amount of antibody while reducing the amount of tracer [4]. This concept is illustrated in Fig. 2.4, which shows an immunoassay calibration curve. The testosterone concentration of a sample is accurately calculated over the dynamic range, represented by the linear portion of the curve. The sensitivity and specificity of the assay is decreased at low testosterone concentrations, where the curve is no longer linear. Changes in the amount of antibody and/or tracer in the assay alter the binding capacity and shifts the curve, thereby affecting the dynamic range (Fig. 2.4).

The RIA requires a specific antibody with minimal cross-reactivity. Testosterone, like most steroids, is a poor antigen. Therefore, the design of a testosterone-specific antibody is, therefore, difficult. For commercially available kits, each manufacturer usually uses a different antibody. This may in turn cause a different binding affinity to testosterone, as well as different cross-reactivity. These differences can certainly contribute to the variable results seen with different commercial kits.

Like RIAs, EIAs are also based on the principle of competitive binding to an antibody. Instead of being radioactively labeled, the tracers used in EIA are fluorescent or chemiluminescent, which eliminates the possibility of radioactive waste. EIAs are easy to automate, and are widely used in non-reference hospital and commercial laboratories.

EIAs are subject to most of the same shortcomings of RIAs, namely, limited utility at very low or very high testosterone concentrations, and the need for a testosterone-specific antibody, which can vary by manufacturer. While these automated assays are sufficient for most clinical purposes in men, results can vary by as much as 33% among different assay methods. Even values in normal men can fall out of the manufacturer’s reference range [14]. Additionally, EIAs do not routinely include a testosterone extraction step. Therefore, they do not necessarily measure true total testosterone.

Some manufacturers of commercial RIA and EIA use danazol for testosterone displacement as an alternative to testosterone extraction using organic solvents. In theory, this additional step is designed to improve the sensitivity and accuracy of the assays. However, the amount of danazol present in the kits is fixed and limited, and usually insufficient in displacing testosterone from samples with elevated amounts of SHBG.

Direct methods of measuring testosterone have gained popularity over the last decade, due to the inconsistencies experienced with indirect immunoassays. Reference laboratories are increasingly using liquid chromatography-tandem mass spectroscopy (LC-MS), which combines automation and high-throughput with high precision, reproducibility, and wide linearity of assay. Steroid hormones are extracted from the serum sample by the use of organic solvents and separated by liquid chromatography, and testosterone levels are then determined by peak area integration of testosterone-containing fractions of the column eluate (Fig. 2.5). LC-MS has a lower limit of detection of 15 ng/dL, with an intra-assay coefficient of variation of <10% [14]. The specificity of this assay is enhanced by a distinct mass spectroscopy fingerprint for all known steroids, from which testosterone can be specifically selected. LC-MS is recognized as the most reliable method of measuring total testosterone in women, children, and hypogonadal men [26, 27], and has been accepted as the new gold standard assay for testosterone measurement.

Unfortunately, there is no cost-effective, simple, widely available, reliable, or accurate method of directly measuring free testosterone. The serum concentration of free testosterone is usually 50–100 times lower than the concentration of total testosterone; therefore, lower limits of detection are important to consider for any assay measuring free T. The free hormone fraction also needs to be separated from bound testosterone in order to be measured.

Equilibrium dialysis is the gold standard assay for the measurement of free testosterone. The patient’s serum is placed in a dialysis chamber with two compartments. One compartment contains 3 H-T (tracer-labeled testosterone) added to the serum, while the other contains a buffer solution. The compartments are separated by a semipermeable membrane with a low molecular weight cut-off. Dialysis is performed until testosterone concentrations within the two compartments reach an equilibrium value. Free and bound 3 H-T are then separated and measured. This value is multiplied by the total testosterone concentration. Although highly precise, equilibrium dialysis methods are expensive and complex, with potential errors due to temperature, sample dilution, and tracer impurities [14]. For example, a 2% contamination, with a tracer that does not bind to proteins can cause a doubling of the free testosterone concentration [5]. The value and accuracy of total testosterone has a bearing on the free testosterone concentration obtained with equilibrium dialysis.

One alternative to equilibrium dialysis for the direct measurement of free T is ultracentrifugation, a process where free T is separated from bound testosterone using ultrafast centrifuges, and then directly measured [29]. The direct measurement of free testosterone by RIA has also been described, but does not provide an accurate assessment when there are alterations in SHBG [30, 31]. The use of this technique is not recommended [32].

Free testosterone can be calculated based on the concentration of SHBG, albumin, and total testosterone. Several different calculations for free T have been described in the literature [33]. These equations require the use of equilibrium dissociation constants for the binding of SHBG and albumin to testosterone, which are usually defined as 1  ×  10⁹ and 3  ×  10⁴ L/M, respectively. These numbers are not universally agreed upon, which can lead to discrepant results when the equations are compared [33]. The albumin concentration in serum is assumed to be 4.3 g/dL for the purposes of this calculation; changes in the albumin level within normal physiologic range have been shown to have little impact on the calculated free testosterone [30]. However, because the calculated value of free T depends on measured SHBG and total testosterone levels, measurement error and variable reference ranges for these two analytes directly impact the results. With the exception of certain metabolic conditions such as obesity, hyperthyroidism, and low cortisol, calculated free T correlates very well with free T obtained by equilibrium dialysis [30]. In contrast, studies comparing calculated values with those obtained by ultrafiltration do not always show high degree of correlation [34].

The FAI represents the ratio of total testosterone and SHBG: testosterone/SHBG. The ratio is easy to calculate, and valid in serum samples from women. There is a reasonable correlation between FAI and FT, but it is not consistently maintained at low testosterone levels [35]. FAI is less useful in men because the majority of SHBG in men is bound to testosterone. Like the calculated free testosterone, the FAI is also dependent on accurate values for testosterone and SHBG.