Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana 46285
Address all correspondence and requests for reprints to: Charmian A. Quigley, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, Indiana 46285.
The postnatal surge of gonadotropins and sex steroids is a physiological phenomenon common to mouse, monkey, and man (as well as rat, horse, and other species; Ref.1). However, since its first description in human male infants in 1973 (2), the mechanisms and physiological significance of this phenomenon in humans have remained a mystery. The study by Bouvattier et al. (3), in this issue of JCEM, evaluates the postnatal hormonal patterns of infants with the androgen insensitivity syndrome (AIS), providing insight into the physiology of the postnatal "mini-puberty" and inviting speculation as to its role in normal postnatal reproductive and sexual development. The data have been long awaited and fill a significant knowledge gap with respect to the endocrinology of AIS. Most striking, and perhaps unexpected, is the distinctive contrast between the postnatal hormonal profile of infants with complete AIS (CAIS), and those with partial forms of the condition. These findings will be of value not only to clinicians who care for infants with disorders of sex differentiation but also to those seeking to better understand the normal physiology of the human reproductive system.
The normal postnatal surge likely results from hormonal events that surround the transition from intrauterine to extrauterine life. The male fetus in late gestation is exposed to high concentrations both of androgens (from testicular secretion) and estrogens (from the placenta). At delivery, the infant is subjected to an abrupt withdrawal of maternal hormones, and his hypothalamic-pituitary-gonadal axis is released from the inhibitory effect of his mothers estrogen. While E2 concentrations are high, serum gonadotropins are low in cord blood, reflecting the final hours of prenatal suppression by maternal estrogens (4). Thereafter, as the inhibitory effect of maternal estrogen wanes, the infants gonadotropins increase significantly, beginning around 1 wk after birth. LH concentrations are within the adolescent range by 12 wk after birth and peak at around 12 months of life, declining slowly thereafter to achieve typical prepubertal childhood values by around 46 months of age (3, 4, 5, 6, 7). In response to the increase in gonadotropins there is a rapid rise in T (and its precursors, 17-hydroxyprogesterone and androstenedione), and the number of testicular Leydig cells increases during third month of life (8, 9, 10). Subsequently, T concentrations decline in parallel with LH (5, 7).
Although the postnatal activity of the LH-Leydig cell axis has been established for some time, the infantile activity of the FSH-Sertoli cell axis has been defined only more recently (11). Like LH, FSH rises soon after birth, although its rise is less dramatic (4, 7). The increase in FSH is accompanied by a rise in the Sertoli cell product, inhibin B, with peaks of both hormones occurring at around 3 months of age (11). FSH subsides to prepubertal values by around 9 months of age, whereas inhibin slowly declines to plateau by about 15 monthsthe last of the hypothalamic-pituitary-gonadal hormones to subside (11). The persistent increase in inhibin concentration presumably reflects ongoing proliferation of Sertoli cells. Because Sertoli cell number is a determinant of spermatogenic potential, the perinatal activation of Sertoli cell proliferation is potentially important for subsequent sperm production in adulthood. Accompanying, and probably stimulated by, the increases in gonadotropin and T secretion, germ cells proliferate during the first 100 postnatal days (12). In addition, there is a T-driven increase in the transformation of gonocytes (primitive reproductive cells) to spermatogonia (9). These changes occur in a developmental window during which gonadotropin stimulation of the testis may be critical to Sertoli cell function and spermatogenesis later in life (13). After 6 months of postnatal life, spermatogonia decline in number as concentrations of gonadotropins and sex steroids wane (12).
The postnatal activation of the hypothalamic-pituitary-gonadal axis also occurs in female infants, but the dynamics are more complex and hormonal levels heterogeneous (1, 4, 7, 8, 11). The gonadotropin surge differs in character between male and female infants. Whereas LH is the dominant hormone in the male surge, FSH predominates in the female (4, 7). Under the influence of the modest postnatal LH rise, E2 secretion is increased during the first 24 months of life (8, 14) in response to the rapid increase in ovarian follicular maturation that occurs during the first 4 months of postnatal life (15). The maturation of ovarian follicles is stimulated by postnatal FSH secretion, which peaks around 36 months after birth. FSH concentrations decline by 12 months, but remain measurable to 24 months. In parallel to the postnatal peak of FSH-stimulated ovarian follicle development, immunoreactive inhibin concentrations in the ovaries (reflecting activity of the granulosa cell component of the ovarian follicle) also peak during the first 4 months (16). The peak of serum inhibin B is variable, occurring between 2 and 12 months, and the decline thereafter is slow and incomplete, with concentrations remaining above prepubertal levels until at least 2 yr of age (11, 14). The FSH-predominant pattern of the postnatal "mini-puberty" in girls forecasts their future development; the earlier rise of FSH than LH secretion is seen again at the onset of true puberty about 10 yr later.
Gonadotropin release by the pituitary gland is controlled primarily by GnRH-secreting neurons in the preoptic area and medial basal hyopthalamus (17). T slows the rate of discharge of the GnRH "pulse generator" in the arcuate nucleus, thereby reducing LH and FSH pulse frequency (18). This effect is also seen with nonaromatizable androgens, indicating a direct action of T, rather than via conversion to E2. The primary site of T negative feedback is the hypothalamus, as evidenced by the retained pituitary response to GnRH (increase in gonadotropin secretion) in T-treated males (19). However, there may also be some direct effect of T at the pituitary level, probably via conversion to dihydrotestosterone (DHT), because there is substantial activity of the 5- reductase enzyme system in the anterior pituitary glands of many species (20). E2, in contrast, exerts its primary effect on the pituitary gonadotrope, resulting in decreased amplitude of LH pulses (19), but also has some feedback effect at the hypothalamic level.
For androgens to exert a direct effect on hypothalamic development and function, ARs should be present in appropriate areas of the hypothalamus and/or pituitary gland. Furthermore, sexually dimorphic expression of ARs might be expected with respect both to distribution and level of AR expression. Indeed, fetal and adult male rhesus monkeys have significant expression of AR in neurons of the medial and ventromedial hypothalamus (21), and a sex difference in AR expression in these hypothalamic regions has been reported in hamsters (22). In the human brain, AR immunoreactivity is generally greater in males than females (23, 24); there are also notable sex differences in brain AR distribution (24). In adult males, AR is primarily localized in the neurons of the caudal hypothalamus and mamillary nuclei. The mamillary bodies are postulated on the basis of animal studies, to be involved in arousal of sexual interest. There are strong sex differences in AR expression in these areas, and the intranuclear localization of AR (where it is required for action as a transcription factor) is also androgen dependent (24). These differences point to a role for androgens and AR in sexually dimorphic hypothalamic processes, such as control of gonadotropin secretion, neural organization, and sexual behavior (24, 25). In addition, a role of androgens in direct regulation of the pituitary gonadotrope is suggested by the finding of AR expression in the pituitary glands of fetal baboons (26), the pituitary gonadotropes of prepubertal male rats (27), and the pituitary glands of adult male rhesus monkeys (28). Expression of ARs in the human pituitary gland has not yet been reported.
The differences in timing, character, and relative intensity of the postnatal surge between male and female infants invite speculation as to the physiological basis and role of this process in each sex. The prenatal male hypothalamus is exposed to a different hormonal milieu from that of the prenatal female. Although both sexes experience high estrogen concentrations, only males are subjected to high-level, high-potency androgens. Because androgens and estrogens impose negative feedback on different aspects of gonadotropin secretion, it seems likely that the differences in postnatal activity of the hypothalamic-pituitary-gonadal axis are a direct consequence of variations in prenatal hormone exposure. Due to the prenatal action of both androgens and estrogens, the infant male hypothalamus has presumably been more tightly restrained than that of the female. The removal of this "double repression" (T feedback at the hypothalamic level and estrogen feedback at the pituitary) may explain the greater vigor of the postnatal surge in male infants compared with that of females.
Evaluation of the normal events of the postnatal hypothalamic-pituitary-gonadal axis has defined the chronology, and to some extent the physiology, of the postnatal surge. However, understanding the physiological relevance of the phenomenon requires the analysis of experimental animal models, or the "experiments of nature" provided by pathological human conditions. Blocking of the pituitary-gonadal axis during the first 4 months of life in monkeys results in lower sperm counts in adulthood, indicating that normal activation of the hypothalamic-pituitary-gonadal axis at this age may be important for subsequent spermatogenesis (12, 29). In addition, postnatal hormone status has been evaluated in infants with hypopituitarism, cryptorchidism, and, now, AIS. The postnatal LH and T surge is absent or blunted in hypopituitary (30, 31) and cryptorchid infants (5). These hormonal disturbances are accompanied by cytological abnormalities in the testes of cryptorchid infants, including reductions in Leydig cell number and the maturational process of transformation of gonocytes to spermatogonia (9).
As an example of human pathology the androgen insensitivity syndrome has provided significant insights into human physiology. Thirty years ago, analysis of the endocrine profile of patients with AIS helped to elucidate the role of sex steroids in regulation of the male hypothalamic-pituitary-gonadal axis. Individuals with the complete form of AIS lack functional ARs in all tissues, including the hypothalamus and pituitary gland. Consequently, these individuals are resistant to androgen negative feedback at the hypothalamic-pituitary level. This is demonstrated by marked elevations of LH, due to increases in pulse amplitude and frequency (32). As a consequence, androstenedione and T are significantly increased in postpubertal individuals with testes in situ (33, 34). Because these individuals have ample serum (and presumably hypothalamic) estrogen concentrations and normal estrogen receptors, their supranormal LH must reflect the requirement for specific androgen action in negative feedback regulation of gonadotropin secretion. These data support the work of Santen (19), who highlighted the role of nonaromatizable androgens in the negative feedback regulation of LH secretion in the normal male. The secretion of FSH and inhibin are essentially normal in individuals with AIS, indicating lack of dependence on androgen action in regulation of this branch of the hypothalamic-pituitary-gonadal axis.
Although studies in postpubertal individuals with AIS have underscored the significance of androgens in adult male hypothalamic-pituitary regulation, little has been reported in prepubertal subjects, and even less in infants. Two prepubertal children with CAIS aged 8 and 11 yr, were reported to have normal LH, FSH, T, and DHT for age (35). In contrast, in an infant with mild (grade 3) partial AIS the concentrations of LH, T (both at baseline and in response to human CG), and androstenedione were increased from 9 d of age (36). Similarly, a premature infant with partial AIS reported by Nagel et al. (37) had normal serum T at 2 d of age, followed by marked elevations of LH, T, and DHT at 17 d of age. Only one longitudinal study from infancy to puberty has been reported. Isurugi et al. (38) followed a single patient with partial AIS from 4 months of age onward. T was high at 4 months (LH and FSH were not measured); thereafter, all hormonal values (LH, FSH, and T) were in the normal range for age on annual analysis performed between 2 and 10 yr of age. After that time, a rapid rise in gonadotropins and T occurred, marking the onset of puberty.
Although hormone concentrations in the few reported infants with partial forms of AIS were elevated in similar fashion to those of postpubertal subjects, no data for infants with CAIS have been reported to date. In the absence of data, it has generally been assumed that the typical hormonal manifestations of postpubertal androgen insensitivity (supranormal LH and T) would be present from birth in patients with CAIS. However, based on our own unpublished observations and those of Prof. J.-L. Chaussain (senior author on the current article), we hypothesized that the postnatal surge did not occur in infants with the complete form of AIS (39). The present study supports this hypothesis. The authors evaluated postnatal gonadotropin and T secretion in the first 3 months of life in 15 infants with AIS due to proven mutations in the AR gene. Whereas the prenatal hypothalamus of the normal male fetus is subject to negative feedback from both testicular androgens and placental estrogens, the hypothalamus of the fetus with AIS is resistant to androgen feedback but retains sensitivity to estrogen. As would be expected in the situation of blunted hypothalamic-pituitary androgen feedback, infants with partial AIS (n = 5) had high postnatal LH and T concentrations, with the peak being at 30 d of age for most of the infants. In contrast, all but one infant with CAIS (n = 10) had low or undetectable LH and T concentrations at 12 months of agethe time of the normal postnatal peak of these hormones in male infants. Both groups of infants had brisk T responses to human CG, indicating normal testicular responsiveness to gonadotropins. However, there was a marked divergence with respect to gonadotropin response to GnRH. The infants with CAIS had blunted LH responses to administered GnRH, whereas those with partial AIS had normal or exaggerated responses.
These important clinical observations indicate that the normal postnatal LH and T surge in human male infants requires hypothalamic expression of an AR that retains at least some degree of function. Furthermore, in infants with partial AIS, the contrasting findings of a retained postnatal surge in the presence of impaired masculinization indicate that the hypothalamus is relatively more sensitive to even a low level of AR action, than the external genital anlagen. The profound absence of the postnatal surge in infants with CAIS in this study was striking. All but one infant showed not even a blip in LH or T concentration. In fact, the gonadotropin concentrations of the infants with CAIS did not achieve even the modest rise typical of the normal female infant. In addition, at 3 months of age the infants with CAIS had a reduced LH response to administered GnRH, likely representing a reduction in the releasable LH stores due to diminished endogenous GnRH tone.
The failure of infants with CAIS to undergo a postnatal surge even comparable with that of the typical 46,XX female infant with ovaries suggests that the gonadotropin and sex steroid surge of normal female infants (albeit of lower intensity than that of the male infant) may be due, at least in part, to withdrawal of androgen (e.g. maternal androgen), rather than estrogen feedback. This implies that hormonal imprinting of the female hypothalamus is an active process, not simply a passive one due to absence of the high level pre- and postnatal androgen exposure experienced by the male fetus. In the absence of pre- and postnatal androgen effect it might be assumed that subsequent function of the hypothalamic-pituitary-gonadal axis, and indeed the rest of the brain, of individuals with CAIS would more closely resemble that of the typical female than that of a typical male. However, this concept remains to be addressed, because no study has examined the longitudinal pattern of gonadotropin secretion in these patients and, other than gender identity and sexual preference typical for females (40), no data are available regarding brain structure or function in individuals with CAIS. The absence of postnatal hypothalamic-pituitary-gonadal activity of the infant with CAIS indicates that in essence, the hypothalamus is neither masculinized nor feminized. It could be suggested that the absence of the postnatal surge in infants with CAIS represents another manifestation of their so-called "superfeminine" statea term originally coined by Lawson Wilkins to highlight the fact that his patient with the condition had even a lesser degree of androgenization than a typical 46,XX female (41). Additional manifestations of this state, such as reduced clitoral size (42) and reduced bone mineralization ( 43), also imply a role for androgens in other aspects of normal female development.
The physiological significance of the postnatal gonadotropin and sex steroid surge has been studied in rodents and primates, but extrapolation from these species to man is speculative. Animal studies suggest roles at three levels of the reproductive systemendocrine, anatomical, and behavioral. The effect of alterations in the postnatal hormonal status on subsequent endocrine function has been evaluated in rodents. T administration to neonatal female rats ablates cyclic hypothalamic GnRH production (44), indicating that the T surge is involved in masculinizing the hypothalamus, programming it to subsequently secrete GnRH in a tonic (male-type) rather than a cyclic (female-type) fashion. However, there is no evidence in man that neonatal deprivation or excess of androgen has any permanent effect on hypothalamic-pituitary function. In fact, two lines of evidence argue against a role for the postnatal surge in hypothalamic masculinization in humans. First, female infants with congenital adrenal hyperplasia who are exposed to masculinizing concentrations of androgens in the pre- or perinatal period nevertheless subsequently develop the typical cyclic female patterns of gonadotropin and sex steroid secretion (45). Second, women with CAIS and testes in situ who, as evidenced by the present study, apparently are not exposed to masculinizing postnatal T concentrations, nevertheless have no evidence of a "cycle," with respect to hypothalamic-pituitary-gonadal function (D. Leary, AISSG, 1 personal communication). That is, the simple absence of androgen effect does not seem to have resulted in a "feminized" hypothalamus. In fact, this observation (which needs to be tested by clinical studies) suggests that the human hypothalamus does not have an intrinsic cyclic pattern and that primary driver of the cyclic activity of the human female hypothalamic-pituitary-ovarian axis is not the hypothalamus, but the ovary itselfpresumably via the process of follicular maturation. Thus, the differentiation of hypothalamic GnRH-secreting neurons toward tonic vs. cyclic function may derive not from the pre- or early postnatal hormonal environment, but instead may develop during puberty or require a sexually dimorphic process independent of hormonal events.
The second proposed role for the postnatal surge is anatomicalthe priming of target tissues for subsequent growth and maturation in later life. In male rats the surge is required for the differentiation and maturation of the accessory sex glands (1). A similar role may exist in primates and humans. Blockade of the neonatal activation of the pituitary-testicular axis in male monkeys results in reduced postnatal penile growth (46) and subnormal testicular enlargement at the time of puberty (29 .); female monkeys exposed to high postnatal androgen concentrations develop clitoromegaly (46). In man, there is some evidence for a role of the T surge in preventing involution of the male external genitalia in the first year of life, as demonstrated by Main et al. (31). Three male infants who failed to undergo the postnatal surge due to hypogonadotropic hypogonadism or hypopituitarism had involution of the penis and ascent of the testes in infancy. The authors hypothesized that the role of the surge is to prevent the regression of the male genitalia. In this regard, T would be acting much as it does during fetal development, when it stabilizes the Wolffian ducts, preventing their regression and promoting their differentiation into epididymis and seminal vesicles. Postnatal androgens may act to prime the male urogenital tract by promoting early growth and preparing for the maturational effects of the sex hormones at puberty. Boys who are born with an unusually small penis undergo normal growth of the external genitalia if T is administered during infancy (47), but may have subnormal genital growth if androgen therapy is delayed until the time of the normal male puberty. In addition to maturational effects on the external genitalia, the postnatal surge also seems to be required for maturation of the gonads themselves in preparation for later reproductive function. The surge is associated with increases in ovarian follicular maturation and in testicular spermatogenesis; absence of the surge, as seen in infants with hypopituitarism and cryptorchidism, is associated with reduced spermatogenesis in later life (9, 12, 29).
The third proposed major role for the postnatal surge is in the imprinting of subsequent sexual orientation and behavior (1). Female rats exposed to androgens in the perinatal period develop sexual behavior typical of males. Female marmoset monkeys develop masculuinized behavior patterns when treated with T in the neonatal period (48), and the rise in T level is postulated to be involved in sexual differentiation of the central nervous system (49). Somewhat controversial evidence for a role in sexual behavior in humans derives from studies of females exposed to excessive pre- and postnatal androgen levels (such as those with congenital adrenal hyperplasia), in whom an increased rate of homosexuality has been reported (50). Similarly, a correlation may exist between the presence of, and a degree of sensitivity to, the T surge in infants with partial AIS, and a subsequent tendency to same-sex sexual orientation in those reared as female. This latter observation requires further study.
In summary, the postnatal gonadotropin (and sex steroid) surge seems to represent the response of an androgen-sensitive hypothalamus (perhaps the arcuate nucleus) and pituitary gland to the withdrawal of androgens, both in male and female infants. These observations imply at least a complementary role of androgens in many processes of normal female development. The role of the postnatal surge in humans does not seem to be the priming of the hypothalamus for later tonic (male) or cyclic (female) GnRH secretion. Rather, human female hypothalamic-pituitary-gonadal cyclicity more likely derives primarily from the ovary itself. The surge seems to be important for postnatal maturation of the gonads, both in males and females, and may be required for stabilization of male external genital development. Whether this phenomenon significantly impacts gender identity and sexual behavior in humans requires additional study.
A number of questions remain to be addressed. Does the presence of a typical to exaggerated male-type postnatal LH and T surge impact later hypothalamic function, gender identity and sexual orientation of individuals with partial AIS? What is the functional nature of the hypothalamus in individuals with CAISis there any cyclic pattern to its hormonal activity? How and when does the hypothalamic-pituitary-gonadal axis become active in patients with CAIS? That is, after its unusual quiescence in infancy, how does the ability of the hypothalamus to "sense" the lack of androgen feedback develop with time? Following the lead of this important work of Bouvattier et al. (5), further clinical studies analyzing the longitudinal secretion of gonadotropins, sex steroids, and other gonadal hormones of individuals with various forms of AIS and other disturbances of sexual differentiation and development are needed.
Acknowledgments
Footnotes
Abbreviations: AIS, Androgen insensitivity syndrome; CAIS, complete AIS; DHT, dihydrotestosterone.
1 Androgen Insensitivity Support Group, http://www.medhelp.org/www/ais/index.htm.
Received October 29, 2001.
Accepted October 30, 2001.
References