Key points

Introduction

Male hypogonadism is defined by the Endocrine Society as a clinical syndrome that results from the inability of the testes to produce physiologic levels of testosterone (T) and a “normal” number of spermatozoa secondary to a dysfunction in the hypothalamic-pituitary-gonadal axis (HPG). The diagnosis of hypogonadism depends on the assessment of the clinical signs and symptoms of hypogonadism and serum T levels assayed on at least 2 different occasions. The symptoms most suggestive are low libido followed by a reduced quality of erections; however, these markers are subjective and nonspecific. The laboratory’s assessment of serum T levels, in contrast, provides an objective measure of the gonadal status that can support or refute clinical signs and symptoms. Nevertheless, serum T levels can vary widely between samples drawn from the same patient and between the various laboratory assay platforms due to a multitude of factors, such as diurnal variation, systemic illnesses, and seasonal variation, as well as assay-specific factors, and must be interpreted with caution.

Current clinical laboratory assay platforms include immunoassays and mass spectrometry (MS). Despite significant advances to improve the accuracy and precision of currently available assays, limited comparability exists between assays at the lower and upper extremes of the T range. Moreover, there is no currently accepted gold standard method of assessment. In this review, the growing significance of T assays in the aging population is highlighted, the indications for the various assays used to assess free and total T are discussed, and the impact of various preanalytical and analytical factors that can influence the results of T assays are analyzed.

Introduction

Male hypogonadism is defined by the Endocrine Society as a clinical syndrome that results from the inability of the testes to produce physiologic levels of testosterone (T) and a “normal” number of spermatozoa secondary to a dysfunction in the hypothalamic-pituitary-gonadal axis (HPG). The diagnosis of hypogonadism depends on the assessment of the clinical signs and symptoms of hypogonadism and serum T levels assayed on at least 2 different occasions. The symptoms most suggestive are low libido followed by a reduced quality of erections; however, these markers are subjective and nonspecific. The laboratory’s assessment of serum T levels, in contrast, provides an objective measure of the gonadal status that can support or refute clinical signs and symptoms. Nevertheless, serum T levels can vary widely between samples drawn from the same patient and between the various laboratory assay platforms due to a multitude of factors, such as diurnal variation, systemic illnesses, and seasonal variation, as well as assay-specific factors, and must be interpreted with caution.

Current clinical laboratory assay platforms include immunoassays and mass spectrometry (MS). Despite significant advances to improve the accuracy and precision of currently available assays, limited comparability exists between assays at the lower and upper extremes of the T range. Moreover, there is no currently accepted gold standard method of assessment. In this review, the growing significance of T assays in the aging population is highlighted, the indications for the various assays used to assess free and total T are discussed, and the impact of various preanalytical and analytical factors that can influence the results of T assays are analyzed.

Prevalence

Late-onset hypogonadism is considered to be a disease of aging. As the population ages, the epidemiologic burden of male hypogonadism is expected to proportionally increase. The World Health Organization estimates that by 2020, the number of people over the age of 65 will for the first time surpass the number of people less than 5 years of age. According to the same estimates, the number of people over the age of 65 will nearly double from a current population of 524 million people to roughly 1.5 billion by 2050. A population-based observational study by Araujo and colleagues estimated the crude prevalence of biochemically confirmed, symptomatic male hypogonadism (total T <300 ng/dL) to be 5.6% in a cohort of 1475 US men aged 30 to 79 and 18.4% in subset of men aged 70 to 79. The investigators projected in this study that in 2025, as many as 6.5 million American men will exhibit symptomatic androgen deficiency. In addition, comorbidities that accumulate as part of aging can also increase the epidemiologic burden of hypogonadism. The Healthy Man Study showed a strong association between comorbidities, such as obesity and ex-smoker status, and T values over repeat measures drawn over a 3-month duration. Based on these trends, a significant increase in utilization of T assays can be expected, and a thorough understanding of the nuances and limitations of the various T assay platforms will be invaluable in the care of the hypogonadal patient.

Diagnosing hypogonadism

Male hypogonadism results from a reduction of androgen levels due to a dysfunction in the HPG axis. The presence and magnitude of symptoms associated with male hypogonadism depend on the concentration of T available to the target organ. Although the total testosterone (TT) concentration is often used as a surrogate for the amount of T available for the target organ, the bioavailable fraction of the TT is a more accurate measure of T concentrations at the tissue level and correlates better with clinical symptoms than TT. T circulates in either a protein-bound or non-protein-bound (free) state. In a healthy adult man, most T circulates in a protein-bound state attached to either albumin ∼50%, sex hormone-binding globulin (SHBG) ∼44%, or cortisol-binding globulin ∼3.5%. The remaining 2% to 3% of T circulates as unbound, free testosterone (FT). The bioavailable fraction of the TT consists of albumin-bound and FT. Unlike the TT level, the bioavailable fraction is not susceptible to SHBG concentration, which can fluctuate with age, thyroid function, drugs and alcohol consumption, and disorders of the pituitary and the liver.

Diagnosing hypogonadism is challenging in the setting of a patient with the signs and symptoms consistent with hypogonadism, borderline normal TT value, and possible alteration of SHBG level. Under these circumstances, the Endocrine Society recommends measuring free or bioavailable testosterone (BioT) levels with an accurate and reliable assay. Several methods have been used to assess FT and BioT levels, such as equilibrium dialysis (FT _D ), direct estimation of serum free T by an analogue ligand immunoassay (FT _A ), ammonium sulfate precipitation of SHBG-bound T, calculation of the free androgen index (FAI) using the formula TT/(SHBG × 100), and multiple algorithms to calculate the free testosterone index (FTI) based on the concentration of albumin and SHBG. Although the FT _D is considered the gold standard, this method is time-consuming and expensive. In contrast, calculation of the FTI using published algorithms, such as the Vermeulen equation, is inexpensive and has been shown to have good correlation with gold standard methods of comparison.

The Vermeulen equation, which is widely used, determines the FT level using the concentration of T bound to albumin and the albumin concentration. The equation operates under the assumption that the albumin concentration is in the physiologic range of 40 to 50 g/L (5.8–7.2 × 10 ⁻⁴ mol/L). There are several shortcomings to this equation. If the albumin concentration is expected to deviate from the physiologic range, Vermeulen and colleagues recommend calculating the albumin concentration in order to account for the reduced amount of albumin-bound T. In addition, the Vermeulen equation assumes that T does not face significant competition from other steroid hormones, such as estradiol (E2) and dihydrotestosterone (DHT), for its binding site on albumin. Supraphysiologic levels of E2 and DHT will confound the calculation, leading to underestimation of the FT.

In a cross-sectional study of 50 men aged 28 to 90, Morley and colleagues compared the results of different tissue-available T assays, including the ammonium sulfate precipitation method, FT _D , FT by ultracentrifugation (FT _U ), FT _A , FTI, and TT, to determine the utility of each assay in the assessment of gonadal status. The investigators compared the week-to-week variability in BioT and TT levels in a subcohort of 16 men with a mean age of 69.3 ± 1.7 years. The investigators also compared the various assays to FT _D , with the best correlation being with FTI (r = 0.807) followed by BioT levels determined using ammonium sulfate precipitation (r = 0.670). When hypogonadism was defined as a TT value of 300 ng/dL or less and compared with BioT levels derived from the ammonium sulfate precipitation assay, 26% of men deemed eugonadal based on TT values were subsequently classified as hypogonadal based on the BioT values, whereas 16% of men classified as hypogonadal using TT values were reassessed as eugonadal based on BioT values. Morley and colleagues also showed considerable variability in both TT and BioT levels over an 8-week period in a subgroup of 16 men.

In men with signs and symptoms of hypogonadism and no suspected alterations in the SHBG level, the TT level has sufficient sensitivity to diagnose hypogonadism. Morris and colleagues demonstrated this in a cross-sectional study of 1072 men who were undergoing coronary angiography. TT, SHBG, and BioT levels (using ammonium sulfate precipitation) were measured, and FAI and FTI levels were calculated. Of the assessed methods, TT was the best predictor of BioT levels in the whole cohort, whereas SHBG and FAI values were the worst predictors of BioT levels. However, when the serum assay methods were assessed for their power to diagnose hypogonadism using the area under the receiver operating curve values, calculated free T outperformed TT (0.75 vs 0.63) in the subgroup of men with an TT greater than 7.5 nmol/L but less than 12 nmol/L. The Endocrine Society Clinical Practice Guidelines, therefore, recommends measuring the morning TT value as an initial diagnostic test and the bioavailable T level in men with low normal TT values.

Reference ranges of testosterone

Despite the increasing use of T assays in research and clinical practice, no universally accepted cutoff value exists for hypogonadism. Although TT values less than 200 ng/dL are considered diagnostic for hypogonadism, levels between 200 and 320 ng/dL are considered equivocal because of the variable presence of symptoms at different cutoff values in different patients and the lack of agreement among the platform assays used. Several preanalytical and analytical factors can alter the specificity and precision of an assay, including technical limitations of the assay and intraindividual variation. This variation is major problem because different assays and laboratories lack comparability in results, especially with lower T values.

In order to address the lack of comparability, the Endocrine Society along with the Centers for Disease Control and Prevention (CDC), National Center for Environmental Health, and Division of Laboratory Sciences have set forth an external quality control program called the CDC Laboratory/Manufacturer Hormone Standardization (CDC HoSt) program to reduce the measurement bias and improve the comparability of T testing methods. The CDC HoSt program provides common calibrators derived from individual donor sera with known target values to align the assay results of participating laboratories. Four sets of 10 serum samples with undisclosed T concentrations are sent to the participating laboratories for analysis of accuracy and performance evaluation. Between 2007 and 2011, utilization of the HoSt program reduced the measurement bias observed among mass spectrometric (MS) methods by approximately 50%.

Variation in testosterone

Differences in TT and FT levels can be due to intraindividual and assay variability. Intraindividual variation, which can be caused by circadian rhythmicity, has a more profound effect on variability.

Intraindividual variation

Both TT and FT exhibit circadian rhythmicity with peaks in the morning and troughs in the evening. This diurnal variation of total and free T is most pronounced in young men and is blunted with increasing age. In a study comparing a cohort of 18 young men (age 21–37 years) to 28 elderly men (age 67–98 year), morning plasma T levels were 33% higher than evening T levels in the younger cohort ( P <.01), but only 8.4% higher in the older cohort ( P >.05). This circadian rhythmicity was also observed by Diver and colleagues, who performed sequential measurements of TT, FT, BioT, and SHBG every 30 minutes for 24 hours in 10 healthy, young men aged 23 to 33 years and 8 healthy men aged 55 to 64 years. Statistically significant variation was seen in all 4 indices of androgen status with a minimum of 43% reduction seen in TT from peak levels observed between 06.00h and 10.00h to nadir levels seen between 18.00h and 22.00h in both the young and middle-aged groups. Although more recent data published by Welliver and colleagues showed in a population of men presenting for erectile dysfunction that the difference between TT levels were statistically significant between 7.00h and 9.00h and 9.00h and 14.00h in men younger than 45 years of age. This difference, however, was not statistically significant in men over the age of 45. The investigators recommended assessing T levels in men younger than 45 as close to 7.00h as possible and before 14.00h in men older than 45 years of age. These findings were corroborated by Crawford and colleagues, who more recently showed no significant difference in TT values drawn between 8.00h and 11.0h, 11.00h and 14.00h, and 20.00h to 8.00h in a population of elderly men with a mean age of 61. The current recommendation of the International Society of Andrology, the International Society for Study of the Aging Male, the European Association of Urology, the European Association of Andrology, and the American Society of Andrology, however, remains for a serum TT level to be obtained between 07.00h and 11.00h.

Moreover, T values can fluctuate from day to day and week to week within the same individual. A study of aging men between 55 and 70 years found TT to vary by 14.8% (range 0.1%–79%) when values were reassessed on 2 different days within the same month. Morley and colleagues showed significant variability of both TT and BioT levels over a course of 8 weeks. Over this course, 8 of 16 men who were eugonadal at one time point were hypogonadal at a second time point. Vermeulen and Verdonck showed a similar degree of variability in a cohort of 169 men aged 40 to 80 with symptomatic benign prostatic hyperplasia observed with 8 serum T level checks over a 50-week period. Although good correlation was observed (r = 0.849) between the first sampling and the 7 subsequent samples, the mean coefficient of variation (CV) was 16.9% ± 8.4% with 9 men demonstrating greater than 25% variation. Some of this variation can be attributed to biological influences, such as heat exposure, physical training, alcohol withdrawal, and medications. Acute and subacute illnesses can also introduce significant variation in T levels; therefore, the Endocrine Society has recommended delaying the evaluation of androgen deficiency until the illness has resolved.

Reportedly, laboratory errors occur in 0.6% of inpatient and 0.04% of outpatient laboratory tests and can introduce significant heterogeneity in assay results. The vast majority of laboratory errors occur in the preanalytical phase with a frequency as high as 84.5%. Common errors that can occur in this phase are incorrect patient identification/mislabeling, inappropriate container/tube used, inappropriate storage or transfer, or inappropriate quantity or quality of specimen. In a systematic review of errors in laboratory medicine, Bonini and colleagues found the highest rate of outpatient laboratory errors was attributable to incorrect labeling of the sample (36.4%) followed by hemolysis of the sample (32.3%). The investigators recommend routine auditing of rejected samples to determine what factors are associated with rejected samples. Centrifuging a sample of blood and delaying its analysis can cause an underestimation of TT values. Attention must also be paid to the temperature of the sample, not to allow it to reach room temperature. In contrast, TT and BioT levels can be determined from serum stored at −70°C for up to 7 years with excellent stability. Other important preanalytical factors to consider are the type of blood collection tube and the properties of the collection tube, including whether it is glass versus plastic. Additives, such as clot activators, can cause a 4-fold increase in TT, and anticoagulant additives, such as EDTA and sodium citrate, can alter SHBG levels and impair FT level determination. Attention must be paid to the above factors in order to minimize the amount of laboratory error introduced and prevent unintended clinical consequences of a spurious laboratory value.