Measurement of total serum testosterone levels using commercially available kits: high degree of between-kit variability

Objective: The measurement of total serum testosterone has an established clinical role in the management of male hypogonadism and female androgen excess disorders. We studied the between-kit variability and precision of six different commercially available testosterone assays and compared them with an established in-house method. Design: Laboratory observational prospective study. Setting: Tertiary university medical center clinical laboratory. Patient(s): Three groups of samples each of men (n = 36) and women (n = 15) who had high, normal, or low levels of sex hormone-binding globulin (SHBG), respectively, were studied. Intervention(s): Individual and pooled (male and female) serum samples were analyzed for total testosterone concentration using six different commercially available assays and one in-house method. Main Outcome Measure(s): The between-kit variability and the effect of the mean (± SD) SHBG level were determined, the results obtained with the use of the kits and the in-house method were compared, and the intraassay variability (i.e., precision) was evaluated. Result(s): Male samples demonstrated a 26.3%-40.8% variance in the results obtained with different kits, which was greatest for samples with the lowest SHBG levels. For female samples, between-kit variability ranged from 57%–115% (average, 77%). The percent deviation of the results obtained with the use of commercial methods from those obtained with the use of our in-house assay was greater for men (mean variance, 194%) than for women (mean variance, 67%). The female pool intraassay coefficient of variation was 3.8% with the use of the in-house method and ranged from 8.9%-21.2% with the use of the commercial kits. The male pool intraassay coefficient of variation was 3.1% with the use of the in-house method and ranged from 3.3%-5.5% with the use of the commercial kits. Conclusion(s): Most commercially available kits for measuring the total serum testosterone level demonstrated significant between-kit variability, which was greatest for female samples. Further, samples with the lowest SHBG levels had the highest between-kit variances. These data strongly suggest that the measurement of total serum testosterone using commercial kits may have limited utility, particularly for the detection of hyperandrogenemia.