Testosterone, the major androgen in men, is necessary for fetal male sexual differentiation, pubertal development, and the maintenance of adult secondary sex characteristics and spermatogenesis. Testosterone also regulates gene expression in most extragenital tissues, including muscle and bone, and the immune system. The testes are the source of more than 95% of the circulating testosterone in men although the adrenal cortex produces large amounts of the testosterone precursor steroids, dehydroepiandrosterone (DHEA) and androstenedione.

Following birth there is a period of activation of the GnRH pulse generator that stimulates testosterone secretion to peak levels of 200-300 ng/dl (7-10.5 nmol/L) at ages 1-2 months with persistent elevation until age 4-6 months 1. Thereafter, a prepubertal CNS brake is applied to GnRH secretion that causes a sharp decline in testosterone to very low levels until puberty begins 2. Kisspeptin appears to be this regulatory factor, although the mechanisms controlling kisspeptin expression have yet not been ascertained 3. In early puberty, LH and testosterone secretion increase dramatically during sleep 4. This difference can be used clinically to evaluate boys with delayed puberty because the rise to a higher testosterone value in the morning than in the evening may precede pubertal testis growth, and indicates that puberty has begun 5. Adult values are usually achieved by age 16 years, and range from 3-10 ng/ml (10-35 nM).

Frequent sampling of peripheral blood in adult men reveals small moment-to-moment fluctuations in testosterone of 10-15% 6. Measurement of testosterone in spermatic venous blood, however, reveals robust episodes of testosterone secretion occuring about once per hr 7. Because pulse frequency is rapid and pulse amplitude is relatively low, only small fluctuations are generally observed in peripheral blood. Therefore a single blood sample is usually an adequate assessment of testosterone production on a given day. The time of day of blood sampling is an important consideration, however, and a blood sample drawn in the morning between 0800 and 1000h is generally recommended. There is a diurnal variation in testosterone in adult men, with highest levels in the early morning, followed by a progressive decline throughout the day, reaching the lowest levels in the evening and during the first few hours of sleep. Nadir values are approximately 15% lower than morning values, although the differences can be as much as 50% 8. There may be a seasonable variation in testosterone as well 9 .

In one study in which multiple blood samples were taken from middle-aged normal men over one year, there was a relatively good correlation (r=0.85) between the plasma testosterone level at the first sampling and the mean value of seven subsequent samples; however, as many as 15% of men were misclassified, with either low or normal values that were not reproducible, so that abnormal and borderline values should be confirmed. 10

Although testosterone deficiency is often an obvious clinical diagnosis, and laboratory tests are merely confirmatory, the diagnosis of hypogonadism is sometimes less straightforward. For most young men, the level of total testosterone in serum remains the best single test to screen for hypogonadism because methodology has been optimized and normative data are widely available. While early assays extracted plasma steroids into organic solvents and separated them by column chromatography, those research methods are too costly for clinical purposes. Commercial assay kits using un-extracted serum or plasma and an 125I-labeled testosterone tracer with highly specific antisera and solid phase separation methods are technically easy to use, precise, relatively inexpensive, and sufficiently accurate for most purposes in adult men 11, 12 . Most kits lack specificity and precision at low values, however, and the results in women and children are much less accurate, and must not be over-interpreted. Results from different laboratories are difficult to compare because between-kit results vary by as much as 2-fold 13.

To reduce personnel costs and eliminate the generation of radioactive waste, automated non-radioactive immunoassay analyzers are now commonly used in hospital laboratories. Quantification is accomplished by fluorescence or electro-chemiluminescence. These platforms use small sample volumes to avoid matrix effects, and support both competitive and proportional two-site formats. While automated assays are probably sufficient for most clinical purposes in men 12 14 careful studies reveal some inaccuracy and imprecision. For example, results in men can vary by as much as 33% among different assay methods, and values in normal men may be less than or exceed the manufacturer’s reference range 15. At the low concentrations found in women and children, results may be 50-100% higher or lower than with RIAs 16 resulting in inaccurate diagnoses. Lipemia may produce high values. Better automated assays are needed 17.

Accordingly, reference laboratories are increasingly employing automated liquid chromatography (LC) tandem mass spectrometry (MS) methodology 16. This approach to high-throughput, non-isotopic assay automation combines the resolution of LC with the specificity of MS, and is now viewed as a gold standard. Larger aliquots of serum are extracted with organic solvents, steroid hormones are derivatized and separated by HPLC, and values are determined by peak area integration of testosterone-containing fractions of the column eluate. LC-MS accurately detects testosterone concentrations as low at 0.5 nmol/L (15 ng/dl) with a within assay coefficient of variation of <10% 18, 19. Thus these assays seem sufficiently accurate for measurements in children and women, although values tend to be lower than with extraction RIA’s. The equipment needed for LC-MS is costly, however, and each LC-MS assay supported by the platform requires substantial analytical development and optimization; personnel must be highly trained. Standardization among laboratories has not been assessed, and comprehensive normal ranges are not yet available.

SHBG is a carbohydrate-rich β-globulin of molecular weight 100,000 K that is produced by the liver and binds testosterone and other steroids, and prolongs their metabolic clearance 20. SHBG had been measured indirectly by radio-ligand binding assays, but sensitive, precise, two site manual immunoradiometric and enzyme-linked assay kits are available, and automated versions have been developed. SHBG is present in adult male plasma in levels that range from 20-100 nmol/L. Between-assay comparisons can be problematic since assay standards vary among manufacturers. In men, SHBG binds 40-60% of the circulating testosterone so that the level of SHBG is one determinant of the total testosterone level. Accordingly, clinical conditions that reduce SHBG levels tend to lower the total testosterone, and those that increase SHBG, increase total testosterone. These conditions are summarized in Table 1. Low SHBG is now recognized as a marker of the Metabolic Syndrome. Men with an atherogenic lipid profile often have low SHBG and low total testosterone levels 21, 22 and low SHBG predicts the development of the Metabolic syndrome and type 2 diabetes mellitus 23, 24. Many studies have shown a negative correlation between BMI, visceral adipose tissue mass or insulin, and SHBG 25. Insulin is known to decrease SHBG gene expression in cultured hepatic tumor cells 26, and lowering insulin levels in women and men with diazoxide increases SHBG 27. Thus, it is likely that hyperinsulinemia is the link between the Metabolic Syndrome and low SHBG, although other factors may be important.

Free and non-SHBG (Bioavailable) testosterone

The level of total testosterone is sometimes inadequate to determine whether testosterone deficiency is present. This occurs with a borderline value, or when the clinical findings and the total testosterone concentration do not agree. This situation occurs most often in obese, type 2 diabetic, and hyperinsulinemic men among whom total testosterone levels are generally low 28, 29 and in older men with total testosterone levels that are within the normal range even though testosterone production declines with aging 30. Under these circumstances, assays for free or bio-available testosterone are more accurate. The bioactivity of circulating testosterone has been alternatively attributed to the small percentage (1-4 %) of total testosterone that circulates unbound (free testosterone), or to the 40-50% of testosterone that circulates unbound or loosely bound to albumin (non-SHBG testosterone or “bioavailable-testosterone”). The dissociation of albumin-bound testosterone is very rapid, and a short dissociation time is thought to allow the albumin-bound fraction to be available for uptake by cells 31.

The free testosterone concentration can be calculated from the product of the total testosterone level and the percentage that is free. The latter can be determined by equilibrium dialysis. This procedure examines the distribution of 3H-testosterone between two compartments, one containing the tracer added to serum, and the second containing buffer or an albumin solution. The compartments are separated by a semi-permeable membrane with a low-molecular weight cut-off. Dialysis is generally performed overnight at 37C by which time the concentrations of 3H-testosterone in the inner and outer compartments reach an equilibrium value. With centrifugal ultrafiltration, dialysis is facilitated and accelerated by centrifugation for 30-60 min. The percentage of counts per minute (cpm) of 3H-testosterone within the inner compartment is determined, and is multiplied by the total testosterone levels. The % cpm in the inner compartment containing the plasma is proportional to the concentration of SHBG. Since the percent free testosterone in men is usually 1-4%, the free testosterone level commonly ranges from 4-20 ng/deciliter (140-695 pmol/L). Equilibrium dialysis methods are costly and complex with potential errors due to temperature, sample dilution and tracer impurities, among other problems. The level of free testosterone in serum can also be measured directly by RIA of the dialysate, avoiding the use of 3H-testosterone, but this approach requires a very sensitive and accurate immunoassay.

A second approach to adjust the total testosterone level for a high or low SHBG concentration is to calculate the free testosterone level from the levels of total testosterone and SHBG, using binding constants for SHBG and albumin. The K _Dfor testosterone for SHBG is most often defined as 1 x 109 L/M and for albumin, 3 x 104 L/M. Differences in the level of albumin in the sample have little impact on the calculated free testosterone, and are ignored. Both the free and non-SHBG bound testosterone can be readily computed using an internet program ( www.issam.ch/freetesto.htm). There is an excellent correlation between the level of free testosterone obtained by equilibrium dialysis and the calculated free testosterone level 32. Because the result is calculated from the testosterone and SHBG levels, measurement error and variable reference ranges for these two analytes impact directly on the measured free and non-SHBG testosterone. Reference ranges vary among assays, and have not been standardized.

The free testosterone index (FAI, or free androgen index) represents the ratio: total testosterone/SHBG (both in units of nmol/L). This adjustment is easy to calculate, and is believed to be valid in serum samples from women. It is less useful in men 33 because most of the SHBG in men is bound to testosterone. Like the calculated free testosterone, the FAI is also dependent on accurate values for testosterone and SHBG.

Non-SHBG-testosterone is called "bio-available" testosterone because adding SHBG to an androgen-containing sample reduces its androgen receptor binding activity 34 and there is evidence in experimental animals that both the free and albumin bound testosterone fractions are biologically active 35. There is a highly positive correlation between the levels of calculated free testosterone and non-SHBG testosterone in plasma (Figure 1). In cross-sectional studies of healthy older men, the results of both assays predict muscle strength and bone mineral density 36, 37. The non-SHBG-testosterone level is generally determined by adding a tracer amount of 3H-testosterone to the serum sample, and selectively precipitating the SHBG-bound 3H-testosterone by adding 50% ammonium sulfate. The 3H-testosterone that remains in the supernatant is presumed to be free or albumin-bound, and is counted. The percentage of counts added that is found in the supernatant is multiplied by the total testosterone concentration in order to determine the non-SHBG testosterone. The assay is technically easy to perform, but is a two-step procedure requiring specialized reagents, and 3H-testosterone, and is not readily automated. Differences in reagent purity can affect the results, and the separation of SHBG from albumin is presumed, but not verified. Furthermore, between- laboratory standardization has not been demonstrated. The non-SHBG testosterone can also be calculated from the level of SHBG and testosterone using the Sodergard equation.

Figure 2. Relation between the percent free testosterone and SHBG in serum from normal men using the analog free testosterone assay by DSL (A) and DPC (B). The free testosterone level (C) was calculated, and the non-SHBG -bound testosterone level was determined (D) in the same samples. The correlation between percent free testosterone or non-SHBG testosterone, and the level of SHBG is shown. Adapted from reference 22.

The direct free testosterone assay was developed as a single-step, non-extraction method in which an 125I-labeled testosterone analog competes with free testosterone in plasma for binding to a testosterone-specific antiserum that has been immobilized on a polypropylene assay tube. The analog has a low affinity for SHBG and for albumin. The values for normal men with this method, as a percentage of the total testosterone (0.2-0.64%), are substantially lower than the 1.0-4.0% determined by other methods. While this difference alone does not cause a problem if adequate reference ranges are available, it immediately prompted speculation concerning the accuracy of the method. Analog assays propose to be unaffected by the effect of a high or a low level of SHBG. Thus it follows that when testosterone production is normal but SHBG is low, the percentage of non-SHBG testosterone in the sample should be higher. On the other hand, when SHBG in a eugonadal man is elevated, the percentage of free testosterone should be reduced. For example, the percentage of free testosterone is inversely proportional to the concentration of SHBG and the percent free testosterone in plasma is inversely correlated with the level of SHBG as men age (21). As shown in Figure 2A and B, however, using two commercially available direct free testosterone assays, the percent free testosterone is unrelated to the level of SHBG in normal men. For comparison, the expected inverse relation between SHBG and the calculated or the non-SHBG-testosterone level is shown for the same samples. Moreover, like total testosterone, free testosterone levels in serum samples from adult men measured with analog methods tend to correlate positively with the level of SHBG 38. Thus, the free testosterone level determined with analog assays appears to provide essentially the same information as the total testosterone level in men, and these assays appear not to correct for differences in the plasma level of SHBG, and are not recommended for this purpose. Many hospital laboratories continue to use these assays, however, although they are slowly disappearing from research articles related to testosterone.

Figure 1. Correlation between the level of non-SHBG testosterone (bio-available testosterone) and the level of free testosterone calculated from the levels of total testosterone and SHBG, in serum samples from normal adult men.

Cell based reporter bioassays have been developed to analyze androgen bioactivity in biological samples. A stable cell line is created by transfection with plasmids encoding the human androgen receptor and a reporter system containing an androgen-responsive gene such as the mouse mammary tumor virus (MMTV)-luciferase reporter. When cells are stimulated with androgens, luciferase activity is increased dose-dependently 39. These assays remain investigational.

The levels of estradiol and estrone in serum are often measured in men with gynecomastia, or with unexplained gonadotropin deficiency because these conditions are occasionally due to estrogen-producing tumors, or to acquired or genetic abnormalities in which estrogen production is increased. The physiological function of estrogens in men is also of substantial research interest 44. Estradiol is produced from testosterone, and estrone is produced from androstenedione, by aromatase P450, the product of the CYP19 gene. This enzyme is expressed in Leydig cells and in the adrenal cortex, as well as in adipose- and skin-stromal cells, aortic smooth muscle cells, kidney, skeletal muscle cells, and the brain. The promoter sequences of the P450 aromatase genes are tissue-specific, but the translated protein appears to be same in all tissues. Increased aromatase expression in adipose and skin stroma with obesity is the most common cause for mild estrogen excess in men. Interestingly, studies of the age-associated decline in testosterone often do not find a parallel fall in plasma estradiol levels perhaps because of increasing aromatase activity and fat mass with aging 45.

Traditional immunoassays for estrogens in men employ relatively large volumes of plasma (2-5 ml) that are extracted with organic solvents. The extract may be subjected to column chromatography in order to separate interfering substances prior to immunoassay, and purified 3H-estradiol is used as the tracer. With those methods, assay sensitivity is approximately 4 pg/ml (15 pmol/L), and typical values in adult men are 15-65 pg/ml (55-240 pmol/L) for estrone, and <10-55 pg/ml (< 200 pmol/L) for estradiol. Those assays are time-consuming and expensive to perform. Instead, commercial kits for the direct assay of estradiol and estrone using 125-I labeled tracers, or nonradioactive methods, that have been optimized for the higher values normally found in pre-menopausal women, are often employed. At low values typical for men, the results of direct assays are less precise, and tend to overestimate true values. Non-radioactive automated methods have also been employed but the very small sample volumes used produce unexpectedly high values. None of the automated multianalyte systems appear to have the functional sensitivity required for the evaluation of estradiol in men 16 The much lower levels of estradiol from ultra-sensitive recombinant cell bioassays suggest that such assays may be needed to better understand the role of estradiol in men 46.

Most of the circulating estradiol in men is loosely bound to albumin or is unbound 47 and only about 20% is bound to SHBG. Moreover, the estradiol that is bound to SHBG binds with lower affinity than does testosterone. Therefore the serum level of SHBG was not predicted to influence the actions of estradiol. In several studies, however, non-SHBG bound estradiol levels correlated more strongly with low bone mineral density and with indexes of high bone turnover in older men than did levels of total estradiol 48. Non-SHBG estradiol can be determined by ammonium sulfate precipitation using 3H-estradiol. Non-SHBG and free estradiol can also be calculated from the total level of estradiol, the level of SHBG, and binding constants of estradiol to SHBG and albumin of 0.3 x 109 L/M, and 3 x 104 L/M, respectively.

FSH and LH, together with TSH and hCG, form a closely related family of heterodimeric glycoprotein hormones. Each hormone consists of a common α-subunit that is non-covalently linked to a specific β-subunit. Both subunits have asparagine-linked carbohydrate chains (2 for human α-subunit, 2 for FSH-β, and 1 for LH-β). The oligosaccharides project from the peptide skeleton, and by shielding of epitopes, and altering the tertiary structure of the hormone, the sugars may impact on antibody binding as well as receptor activation and bioactivity. Glycosylation also prolongs hormone clearance.

Highly sensitive and specific two site assays for LH and FSH are available. Specificity results from the use of two distinct monoclonal antibodies; often one antibody is to the α-subunit, and the second antibody is to the β-subunit. One antibody is immobilized and is known as the capture antibody. Binding of hormone to the second, or detection antibody, produces a signal that is proportional to the level of hormone present.

Because LH is released from the pituitary into the circulation in pulses, the range of normal values is quite broad. Figure 3 shows that basal LH and FSH values rise slowly during puberty, and that there is substantial overlap among the various pubertal stages 49. Values in adult men ranged from 1.4 to 9.2 IU/L. Two site LH assays are sufficiently sensitive to detect pulsatile LH secretion in prepubertal children 50 and are hormone-specific and unaffected by the cross-reactivity with uncombined gonadotropin α-subunit that occurred with first generation double-antibody polyclonal assays. That effect produced abnormally high LH values in samples from patients with α-subunit producing pituitary tumors or patients with chronic renal failure who have reduced α-subunit clearance. Early LH immunoassays also cross-reacted with hCG, as in samples from pregnant women or cancer patients who produce this gonadotropin ectopically.

Figure 3. Serum LH and FSH levels in normal males measured by AutoDELFIA assays. Redrawn from Brito et al. J Clin Endocrinol Metab 84:3539, 1999.

Variable pituitary and recombinant preparations used as assay standards can make it difficult to compare quantitatively the results from different laboratories 51. Although the number of CHO chains is constant for each subunit, differences in sugar sequences and branch patterns occur in assay standards, and some of the variation between laboratories can be accounted for by variable glycosylation of the standards 52. For example, rh-LH produced by CHO cells is sialylated only in an A2-3 confirmation, and sulfated terminal glycosylation is absent 53.

Two-site assays can occasionally be "too specific". Because of the presence of polymorphisms that impact on immunoassay detection by monoclonal antibodies, misdiagnoses could occur. For example, there is a relatively common polymorphism in the LH-β gene that is characterized by two point mutations in codons 8 and 15 resulting in two amino acid substitutions and an extra glycosylation site 54. Even though men and women with the LH variant appear to be normal and fertile, the serum LH level was diminished or absent using certain monoclonal antibodies, whereas the result with other assays was normal. When a low or undetectable result, or a disparity between LH and FSH levels cannot be readily explained, a second assay method should be used.

Various glycoforms of LH and FSH are found in the pituitary and in the circulation. Highly sialylated glycoforms, with an acidic pH, tend to have a longer circulating half-life, whereas the more alkaline forms tend to exhibit greater bioactivity in vitro. The finding that LH and FSH glycosylation is under physiological control increases further the interest in the gonadotropin glycoforms 55. The ideal immunoassay would detect accurately only the total bioactive LH or FSH in the sample, but clearly this goal is difficult to accomplish. As an alternative approach, in vitro bioassays for LH and FSH have been used to assess the function of the gonadotropins. An in vitro bioassay based on the production of testosterone by cultured mouse or rat Leydig cells 56 is often used to assess LH function. Production of estradiol by rat granulosa cells or immature Sertoli cells has been used to assay for FSH bioactivity 57. While useful for the study of the biological properties of recombinant or purified proteins, the clinical use of these assays can be limited by the nonspecific effects of serum. In fact, many of the findings reported with these assays were found subsequently to be methodological artifacts. Bioassays based on cAMP production by cell lines stably expressing gonadotropin receptors, with quantification using cAMP-responsive promoters linked to a luciferase reporter, have also been developed 58; but these assays remain research tools.

Inhibin, a glycoprotein hormone produced by the testes as well as the ovaries, is responsible for the selective negative feedback control of FSH secretion, and may function as an intratesticular regulator. Inhibin decreases FSH-β mRNA levels by blocking activin signaling 65. Inhibin is a 32-kDa heterodimer composed of an a, and one of two β subunits, β _Aor β _B. Inhibin-B (α-β _B) is the form produced by testicular Sertoli cells, whereas inhibin-A, a product of the corpus luteum and placenta, is undetectable in the circulation in men66. Inhibin α-subunit is produced and released into the circulation in excess of dimeric inhibin-B. Early double antibody immunoassays detected both the bioactive dimer and the seemingly inactive uncombined α-subunit, and therefore did not identify the decrease in inhibin secretion in hypogonadal men with elevated FSH levels.

Several specific two site assays to measure serum inhibin-B levels are available. In one popular ELISA assay, microtiter plates are coated with an antibody to inhibin β _Bthat functions as the capture antibody. The indicator antibody is specific for the inhibin α-subunit, and is coupled to alkaline phosphatase. Values in normal men range from 100-500 pg/ml. 67

The regulation of inhibin-B production is complex, and is incompletely understood. Serum inhibin levels are partly determined by gonadotropin stimulation 68, but are also related to Sertoli cell number 69. Damage to germ cells, with anatomical preservation of Sertoli cells, as occurs following chemotherapy or testicular irradiation, results in low levels of inhibin-B 70, 71. Thus Sertoli cell number, circulating gonadotropins and a germ cell factor all influence inhibin-B production.

Although there is no remarkable pulsatile fluctuation in circulating inhibin-B levels, there is a diurnal variation in normal adult men with the highest values in the morning, and nadir values are approximately 35% lower in the early evening 72. This diurnal pattern of secretion parallels that of testosterone. Inhibin-B levels are measurable in serum in the male fetus and at term, and increase in newborn boys coincident with the rise in gonadotropins and testosterone. Values remain elevated for 2-4 months and then decline 73, 74. But in contrast to the barely detectable levels of gonadotropins and testosterone, circulating inhibin-B levels in prepubertal boys are readily measurable. These findings indicate that inhibin-B production is partly gonadotropin independent. Serum inhibin-B levels increase to adult values at the time of puberty74, and decline as men age 75.

Inhibin levels are reduced in gonadotropin-deficient men 76 with lower values in men with the complete form of congenital hypogonadotropic hypogonadism than in men with partial gonadotropin deficency and some spontaneous pubertal development. This quantitative difference is partly due to a larger mass of Sertoli cells (as reflected by larger testicular size) in the latter group of men. Inhibin-B levels are also low in men with testicular failure. Values may be undetectable in men with Klinefelter's syndrome77, in whom germ cells are generally absent and many tubules are hyalinised, as well as in men with germinal aplasia in whom Sertoli cells are present78. These results imply a role for germ cells in activin/inhibin β _Bgene expression. There is a demonstrable, albeit weak, positive relationship between inhibin-B levels and sperm count 79, and with the germ cell score in testicular biopsy specimens among infertile men 80. Although mean inhibin-B levels are lower than normal in men with impaired spermatogenesis, as shown in Figure 4 there is substantial overlap between values in normospermic and oligospermic men 81. Serum inhibin levels changed little in normal men who developed azoospermia or severe oligospermia in a male contraceptive clinical trial of testosterone together with a progestin 82. Moreover, inhibin B did not predict the presence of spermatozoa, or successful fertilization after ICSI following testicular sperm extraction in azoospermic men 83. Thus the clinical usefulness of inhibin-B assays to measure fertility potential remains to be determined. Further, very large doses of FSH are needed to increase circulating inhibin-B levels in men, so that an FSH stimulation test is likewise not helpful in a clinical evaluation of hypospermatogenesis

Figure 4. Serum inhibin B levels in subfertile men. From Pierik, JH et al, JCEM, 1998.

Inhibin-B is undetectable in most boys with congenital anorchia, as in castrates, and is therefore a useful test to help distinguish these patients from boys with intra-abdominal testes 84

Inhibin- α-subunit is sometimes expressed by testicular, adrenal and ovarian tumors 85-88. There are case reports of a boy with Peutz-Jeghers syndrome and a Sertoli cell tumor in whom the levels of inhibin-B and pro-αC were increased 89 a boy with McCune Albright syndrome in whom G _sα gene mutation in the testes produced autonomous function of Sertoli cells and macroorchidism with increased inhibin-B 90 and an adult man with an adrenal neoplasm that produced inhibin-B 91.

Current inhibin-B ELISAs detect not only 31K inhibin, but also larger (55-100 K) molecular weight forms 92, 93. The bioactivity and significance of the larger forms are not known. In monkeys, differences in 90-100 K inhibin-B levels among prepubertal, juvenile and adult animals paralleled those of 31 K inhibin-B. The detection of multiple forms of inhibin-B could partly explain overlap in values among normal and oligospermic men. Accordingly, more specific assays for 31K inhibin-B are being developed.

Mullerian inhibitory hormone (anti-mullerian hormone) is a member of the TGF-β family of growth and differentiation factors. It is produced by Sertoli cells and causes regression of mullerian structures during fetal male development 94. MIH is readily detectable in the serum of prepubertal boys in concentrations of 10-70 ng/ml, but declines to levels of 2-5 ng/ml with entry into adolescence 95. The hormone is absent from the plasma of most prepubertal boys with congenital anorchia, but is generally detectable in boys with bilateral cryptorchidism. Therefore as shown in Figure 5 measuring MIH is useful in the evaluation of boys with non-palpable gonads 95, 96.

Figure 5. Serum Müllerian inhibiting substance concentrations as a function of age in children with abnormal testes and those with normal testes. Lee et al. NEJM 336 (21): 1480, 1997.