0021-972X/00/$03.

00/0 The Journal of Clinical Endocrinology & Metabolism Copyright 2000 by The Endocrine Society

Vol. 85, No. 4 Printed in U.S.A.

Serum Inhibin A and Inhibin B in Healthy Prepubertal, Pubertal, and Adolescent Girls and Adult Women: Relation to Age, Stage of Puberty, Menstrual Cycle, Follicle-Stimulating Hormone, Luteinizing Hormone, and Estradiol Levels*
ASTRID SEHESTED, ANDERS JUUL, ANNA MARIA ANDERSSON, LLER, JRGEN HOLM PETERSEN, TINA KOLD JENSEN, JRN MU NIELS E. SKAKKEBAEK
AND

Department of Growth and Reproduction, Rigshospitalet (A.S., A.J., A.M.A., T.K.J., J.M., N.E.S.), and Department of Biostatistics, Panum Institute (J.H.P.), University of Copenhagen, 2100 Copenhagen, Denmark; and Department of Epidemiology and Public Health (T.K.J.), Imperial College of Medicine at St. Marys, London, United Kingdom W21PG
ABSTRACT Biochemical assessment of gonadal function during maturation in girls and in adult women can be troublesome. With the recent advent of specific assays for the gonadal peptides inhibin A and inhibin B, it might be possible to achieve a clearer picture of events. We therefore determined serum levels of inhibin A, inhibin B, FSH, LH and estradiol in a cross-sectional study of 403 healthy schoolgirls (aged 6 20 yr) in relation to age and stage of puberty and in 181 healthy nonpregnant women (aged 20 32 yr) in relation to stage of the menstrual cycle. In addition, inhibin A and inhibin B were measured daily throughout the menstrual cycle in 10 healthy adult women. Levels of inhibin B are low or undetectable in prepubertal girls (median, 26.5 pg/mL; 95% prediction interval, 20 100 pg/mL), increase sharply through pubertal stage II to peak in stage III (median, 84 pg/mL; 95% prediction interval, 28 227 pg/mL) and thereafter decline through pubertal stages IV and V. These changes presumably reflect increasing ovarian stimulation through early puberty, resulting in an increased number of developing follicles, follicles reaching a later stage of development before undergoing atresia, or both. Declining levels in late puberty and adulthood probably reflect the onset of the menstrual cycle and the subsequent appearance of the luteal phase, where inhibin B levels are low. Inhibin A levels are undetectable or very low in early puberty (median, 7 pg/mL; 95% prediction interval, 714) pg/mL), increasing gradually through pubertal stages to reach their highest values in adult women (median, 21.5 pg/mL; 95% prediction interval, 7129 pg/mL). Levels of inhibin A greater than 19 pg/mL are only seen in postmenarcheal girls in puberty and in adult women, again consistent with inhibin A being primarily produced by the corpus luteum. Determining cut-off levels of serum inhibin B regarding whether a girl had entered puberty resulted in similar (low) sensitivities and specificities as those found for cut-off levels of LH or estradiol due to the large overlap between serum values in Tanner stages I and II. Correlations between inhibin A and inhibin B and FSH, LH, and estradiol within pubertal stages are presented. In early puberty both inhibin A and inhibin B correlated positively with LH and FSH. In late puberty inhibin A correlated negatively with FSH and did not correlate with LH; inhibin B still correlated positively with both FSH and LH, now most strongly with FSH. In adult women during the menstrual cycle, serum inhibin B levels increased during the follicular phase, indicating the greatest production by follicles in early stages of development. In contrast, serum inhibin A levels peaked during the luteal phase, indicating the greatest production by the corpus luteum. In conclusion, serum inhibin A and inhibin B levels in normal puberty in girls show consistency with our knowledge of the manner in which these hormones are secreted within the menstrual cycle in adult women. The presented reference values may be of use in the clinical evaluation of pubertal development in girls. (J Clin Endocrinol Metab 85: 1634 1640, 2000)

RIGINALLY defined by its FSH-suppressing properties (1), the gonadally produced glycoprotein hormone inhibin has since been found to exist in several different biologically active and inactive forms (2). Circulating biologically active inhibin exists as heterodimers consisting of an -subunit disulfide linked to either a A-subunit (inhibin A) or a Bsubunit (inhibin B) (3). Earlier serum inhibin RIAs could not
Received September 2, 1999. Revision received December 3, 1999. Accepted December 17, 1999. Address all correspondence and requests for reprints to: Dr. Astrid Sehested, Department of Growth and Reproduction GR 5064, Rigshospitalet, Blegdamsvej 9, 2100 Copenhagen , Denmark. E-mail: sehested@rh.dk. *This study was supported by the Danish Medical Research Council (Grants 9700833 and 9600821).

distinguish among inhibin A, inhibin B, and inactive free subunits and precursor forms. New two-site enzyme-linked immunosorbent assays specific for dimeric inhibin A and inhibin B have made it possible to begin to elucidate the role these hormones play in gonadal function and the regulation of the hypothalamic-pituitary-gonadal axis in both sexes. In infant boys, levels of inhibin B peak in the supra-adult range at 3 months of age and thereafter decline slowly during the first 2 yr of life (4). In boys, inhibin B levels increase sharply from childhood to adult levels between pubertal stages I and II, and there seems to be a switch in the regulation of inhibin B production between early and midpuberty (5); the negative correlation between FSH and inhibin B seen in adult men (6) first appears in midpuberty. Inhibin A is undetectable in male serum.

1634

SERUM INHIBIN A AND B IN GIRLS AND WOMEN

1635

We have previously shown that inhibin B levels in infant girls also peaked at 3 months of age, although with a lower peak and greater variation than in boys, indicating early activation of the pituitary-gonadal axis. We subsequently found very low or undetectable levels in girls from 6 24 months of age (4). Levels of circulating gonadotropins and sex steroids are low in prepubertal girls and increase through puberty until the adult pattern of cyclical variation in gonadotropin and sex steroid secretion is achieved. Previous studies of total inhibin (i.e. using a RIA assay with overlapping specificity for inhibin A, inhibin B, inactive subunits, and precursor proteins) in girls in puberty have shown that levels increase through puberty (7, 8), but have failed to demonstrate correlations between mean FSH levels and mean inhibin levels. In both studies lower mean inhibin concentrations were found in girls than in boys at all stages of puberty. Using newly developed specific assays, we here present serum values of inhibin A and inhibin B in relation to age, pubertal stage, and serum FSH, LH, and estradiol in a crosssectional study of 403 healthy prepubertal, pubertal, and adolescent females, aged 6 20 yr. Hormone values were also measured in 181 women, aged 20 32 yr, as well as in 10 healthy women throughout their menstrual cycles.
Subjects and Methods Subjects
Girls 6 20 yr. Serum was obtained from 403 healthy schoolgirls. None had acute or chronic disease, and none was taking any medication. Girls taking the contraceptive pill were excluded. Tanner breast stage (9) was assessed by the same two physicians and was available for 402 subjects. The girls were also asked whether they had experienced menarche. All participants and their parents (6 18 yr) gave their informed consent. Other aspects of this study have been presented previously (10). Adult women. Blood samples were taken from 181 healthy nonpregnant women, aged 20 32 yr. The samples were taken at random within the menstrual cycle; the women were taking no medication. Ascertainment of nonpregnant status at the time of blood sampling was achieved by measurement of urinary hCG during the cycle. Other aspects of this study have been presented previously (11). Menstrual cycle. Daily blood samples were drawn from 10 healthy nonpregnant women (aged 18 40 yr) for 1 month. Other aspects of this study have been reported previously (12). All studies had been approved by the local ethics committees.

of variation was less than 8%, and the interassay coefficient of variation was less than 13%. Progesterone was only measured in the 10 women with daily blood samples throughout the menstrual cycle. The analysis was performed by time-resolved immunofluorometric assay, with a detection limit of 0.8 nmol/L, an interassay variation of 10.1%, and an intraassay variation of 7.3%.

Statistics
Reference curves for the highly skewed distribution of the reproductive hormones as a function of age were calculated. The curves represent the median, the 2.5th percentile, and the 97.5th percentile, corresponding to the mean 2 sd and the mean 2 sd, respectively. The reference curves were obtained by locally weighted regression quantiles (15). This allowed taking into account both the age-dependent and highly skewed distribution of the hormones as well as the left-censoring, i.e. the measurements that were below the detection limit of the assays. Hormone values through different stages of puberty in girls and phase of the menstrual cycle in adult women were first analyzed with the Kruskal-Wallis test, and values at each stage or age were thereafter compared with those from the previous stage using the Mann-Whitney U test. Correlations within stages of puberty corrected for age trend and left-censoring were calculated by first subjecting the reproductive hormones to a cubic root transformation. The transformation corrected the marked right skewness of the data and resulted in the best approximation to a normal distribution. Pearson correlations could now be calculated based on a Tobit analysis (16). The analysis was carried out using the procedure LIFEREG in the statistical package SAS (SAS Institute, Inc., Cary, NC). Sensitivities and specificities for various cut-off values of inhibin B, LH, and estradiol with respect to whether the child was clinically in puberty (i.e. in Tanner stage I or in Tanner stage II or II/III) were calculated as described by Cavallo et al. (17). The specificity refers to the true negative fraction, i.e. the proportion of girls clinically in Tanner stage I with hormone values below the cut-off level. The sensitivity refers to the true positive fraction, i.e. the proportion of girls in Tanner stage II or in stage II/III with hormone values at or above the cut-off level. Serum values in adult women. The group of 181 women was subdivided according to phase of menstrual cycle as follows, using the assumption that the luteal phase is of a fixed length of 14 days: early follicular phase, from the first day of menstruation until 22 days before the first day of the next menstruation (day 22) inclusively; late follicular phase, days 21 to 16 inclusively; periovulatory phase, days 15 to 12 inclusively; midluteal phase, days 11 to 6 inclusively; and end-luteal phase, days 5 to 1 inclusively. The median cycle length was 29 days, with a range of 20 68 days. Excluding those women with a cycle length of less than 24 days or more than 35 days made no difference to the results, so they were included. Medians and 95% prediction intervals of reproductive hormones were calculated for each cycle phase. Menstrual cycle. All 10 women had ovulatory cycles, as demonstrated by a LH peak followed by luteal values of serum progesterone. The menstrual cycles were aligned to the LH peak. Mean and sem were calculated for daily measurements of inhibin A, inhibin B, FSH, and LH. Additionally, the mean and sem of the hormone values relative to the intercycle FSH peak as well as relative to the first day of menses were calculated to show the luteal-follicular transition.

Methods
Blood samples were drawn from an antecubital vein between 0800 1300 h and centrifuged. Serum was stored at 20C until analysis. Serum inhibin A and inhibin B were measured in duplicate in double antibody enzyme immunometric assays using monoclonal antibodies raised against either the A- or B-subunit, respectively, in combination with a labeled antibody raised against the inhibin -subunit as previously described (13, 14). The detection limits were 7 and 20 pg/mL for inhibin A and inhibin B, respectively. Intra- and interassay coefficients of variation were, respectively, less than 15% and less than 16% for inhibin A and less than 15% and less than 18% for inhibin B. Serum FSH and LH were measured by time-resolved immunofluorometric assay (DELFIA, Wallac, Inc., Turku, Finland), with detection limits of 0.06 and 0.05 U/L, respectively. Intra- and interassay coefficients of variation were both below 8% in the FSH and LH assays. Estradiol was measured by RIA [Coat-a-Count, Diagnostic Products (Los Angeles, CA) and ImmunoDiagnostic Systems (Boldon, UK), respectively]. The detection limit was 18 pmol/L, the intraassay coefficient

Results Serum inhibin A, inhibin B, FSH, LH, and estradiol levels in relation to age and stage of puberty

The median and 2.5th and 97.5th percentiles of values of inhibin A, inhibin B, FSH, LH, and estradiol in relation to stage of puberty are given in Table 1, and the medians are plotted without percentiles in Fig. 1. The values in 181 adult women are also shown in relation to phase within the menstrual cycle in Table 2. The individual serum inhibin A and

1636

SEHESTED ET AL.

JCE & M 2000 Vol 85 No 4

TABLE 1. Median and 95% prediction interval (2.597.5 percentiles) of serum inhibin A, inhibin B, FSH, LH, and estradiol at each stage of puberty in girls and young adult women
Schoolgirls: stage of puberty I (n 124) II (n 40) III (n 29) IV (n 39) V (n 170) Adult women (n 181)

Age (yr) 2.597.5 percentiles Inhibin A (pg/mL) 2.597.5 percentiles % below detection limit Inhibin B (pg/mL) 2.597.5 percentiles % below detection limit FSH (U/L) 2.597.5 percentiles LH (U/L) 2.597.5 percentiles Estradiol (pmol/L) 2.597.5 percentiles

9.1 5.9 11.8 7 7 to 7 99.2 26.5 20 to 100 35.5 1.43 0.31 4.75 0.05 0.05 to 0.68 22 18 to 52

11.4 9.213.6 7a 7 to 14 90 51b 20 to 240 10 2.98b 0.61 6.84 0.275b 0.05 to 4.31 40.5b 18 to 290

12.2 10.515.9 7a 7 to 52 58.6 84c 28 227 0 4.61b 2.91 8.79 2.16b 0.619.8 128.5b 18 to 1379

14.5 11.517.4 8 7 to 69 48.7 94 20 to 205 12.8 5.07 0.14 8.78 4.11a 0.05 to 20.02 162 18 to 1094

16.85 12.8 19.5 11.5a 7 to 126 28 75d 20 to 177 16.8 4.45c 0.19 9.2 3.77 0.05 to 20 182d 271108

26.6 21.731.9 21.5b 7 to 129 13.5 49b 20 to 185 32.6 4.24 1.6110.52 5.85b 1.96 26.63 289b 74 1075

The percentage below the detection limit of the assay is given for inhibin A and inhibin B. Significance is indicated when the levels are statistically significantly different from the levels of the previous stage of puberty (by Mann-Whitney U test; not shown for age). a P 0.005. b P 0.0005. c P 0.05. d P 0.05, Kruskal-Wallis Tanner stages IIIV.

FIG. 1. Median of serum inhibin A, inhibin B, FSH, LH, and estradiol plotted in relation to stage of puberty or age (20- to 32-yr-old women).

inhibin B levels in girls in puberty are shown in relation to age in Fig. 2. Serum levels of inhibin A, inhibin B, FSH, LH, and estra-

diol all increased significantly between pubertal stages I and II and stages II and III. Inhibin A continued to increase through stages IV and V, with a significant increase between stage V and levels in adult women. In contrast, inhibin B levels showed a significant declining tendency through stages IIIV (by Kruskal-Wallis test, P 0.05), with a further significant decline from stage V in adolescent girls to levels in adult women. FSH values remained at similar levels in stages III, IV, V, and adult women. LH levels continued to increase significantly from stage III to stage IV, remained at similar level in stage V, and thereafter increased significantly in adult women. Estradiol showed a significant increasing tendency through stages IIIV (by Kruskal-Wallis test, P 0.05), with a significant increase between stage V in adolescent girls and adult women. Our group of prepubertal (Tanner stage I) girls consisted of 124 girls in the age range 6 12 yr. Inhibin B, FSH, LH, and estradiol all increased significantly within Tanner stage I as a function of age (Kruskal-Wallis P values over age groups: P 0.0005, P 0.0005, P 0.003, and P 0.004, respectively). Median inhibin B values increased from less than 20 pg/mL in girls younger than 8 yr of age to 38 pg/mL in girls aged 11 yr or older who were still in Tanner stage I. The distribution of values of inhibin A and inhibin B within Tanner stages was highly left-skewed for all stages except inhibin B in stage III, where the distribution was near normal. This was in part due to a proportion of subjects within each stage having inhibin levels that were below the assays detection limits, as described in Table 3. The percentage of subjects with undetectable levels of inhibin A fell progressively through Tanner stages IV and into adulthood, whereas the percentage of subjects with undetectable levels of inhibin B fell from 35% in stage I to zero in stage III and increased to 34% in adult women. No girls in Tanner breast stage I or II had experienced

SERUM INHIBIN A AND B IN GIRLS AND WOMEN

1637

TABLE 2. Median and 95% prediction interval (2.597.5 percentiles) of serum values of inhibin A, inhibin B, FSH, LH, and estradiol in 181 women according to phase of the menstrual cycle (see Subjects and Methods)
Early follicular (n 44) Late follicular (n 41) Periovulatory (n 22) Midluteal (n 49) End luteal (n 25)

Inhibin A (pg/mL) Inhibin B (pg/mL) FSH (U/mL) LH (U/mL) Estradiol (pmol/L)

9 (7 to 34)a 103 (20 to 261)b 6.39 (2.629.91)b 5.93 (2.019.9)c 164 (74 611)

22 (7 to 99)b 82 (20 to 286)a 4.91 (1.9518)a 6.28 (2.134.2) 425 (1131509)b

43 (8 233)b 26.5 (20 to 189)c 4.54 (2.66 16.10) 8.51 (2.9231.1) 330 (171955)

52 (7 to 145) 20 (20 to 164)c 2.94 (1.42 6.23)b 5.52 (1.40 16.28)a 436 (751025)

15.5 (7 to 145)b 20 (20 to 107) 3.23 (0.057.02) 4.32 (0.04 8.29)a 250 (18 to 864)b

Significance (determined by Mann-Whitney U test) is indicated when levels are significantly different from levels in the previous cycle phase. a P 0.05. b P 0.0005. c P 0.005.

the menarche, which was mainly due to an increasingly left-skewed distribution (only 1 of 34 premenarcheal girls in stages IIIV had undetectable inhibin B, whereas 32 of 201 postmenarcheal girls had undetectable levels).
Correlations

The correlations of serum inhibin A and inhibin B to FSH, LH, and estradiol levels according to stage of puberty corrected for age and left-censoring are shown in Table 4.
Sensitivities and specificities of cut-off values of LH, estradiol, and inhibin B in relation to puberty

Sensitivities and specificities for various cut-off levels of LH, estradiol, inhibin A, and inhibin B or combinations thereof were determined in relation to whether the subject was clinically in puberty (both stage II alone and stage II together with stage III; see Subjects and Methods) and are presented in Table 5.
Levels in adult women during the menstrual cycle

FIG. 2. Serum levels of inhibin A and inhibin B in 399 schoolgirls plotted in relation to age. The lines represent the 97.5th percentile, the median, and the detection limit of the assay (7 pg/mL for inhibin A and 20 pg/mL for inhibin B). Previously reported (4) normal ranges of inhibin B for girls aged 0 2 yr are included in the figure.

menarche. Five of 28 girls in Tanner stage III, 29 of 39 in Tanner stage IV, and 169 of 170 in Tanner stage V had experienced menarche. There was a highly significant difference between levels of inhibin A in pre- and postmenarcheal girls in Tanner breast stages IIIV (Table 3). Inhibin B displayed a nonsignificant (P 0.132) tendency to fall after

Serum inhibin A and inhibin B levels in 181 adult women in relation to cycle phase (Table 2, random sample group) and in 10 women who had serum values measured daily through the menstrual cycle with values aligned to the LH peak (Fig. 3, menstrual cycle group) showed similar patterns that were in agreement with previous reports (13, 14, 18, 19). Serum inhibin B levels in the random sample group were at their highest in the early and late follicular phases, fell in the periovulatory phase, and were at their lowest in the mid- and end-luteal phases. Serum inhibin A levels were at their lowest in the early follicular phase, increased significantly in the late follicular phase, continued to increase through the periovulatory phase, were at their highest in the midluteal phase, and fell significantly in the end-luteal phase. The median and 95% prediction interval according to cycle phase in the menstrual cycle group were calculated for comparison (not shown). The numbers were similar, with two differences: 1) there were no significant differences in inhibin B levels among the early follicular, late follicular, and periovulatory phases in the menstrual cycle group; and 2) there were women with unmeasurable inhibin A levels in the periovulatory and midluteal phases in the random sample group. This was not the case in the menstrual cycle group, where the lowest measured serum inhibin A levels at any time during the periovulatory and midluteal phases were 16 and 23 pg/

1638

SEHESTED ET AL.

JCE & M 2000 Vol 85 No 4

TABLE 3. Serum values of inhibin A, inhibin B, FSH, and LH in pubertal girls in Tanner stages IIIV before and after menarche
Girls in breast stage III, IV, or V Age Inhibin A Inhibin B FSH LH

Premenarcheal (n 34), III/IV/V:23/10/1 Postmenarcheal (n 203), III/IV/V:5/29/169

12.6 (10.516) 16.6 (11.520.8)

7 (7 to 19) 11 (7 to 159)a

80 (20 to 227) 79 (20 322)b

4.82 (1.24 8.79) 4.58 (0.06 to 12.68)c

2.26 (0.16 7.34) 3.95 (0.05 to 74.17)a

Values are given as median (range). a P 0.0005 for difference between pre- and postmenarcheal girls. b P 0.132. c P 0.123. TABLE 4. Correlations between serum inhibin A and serum inhibin B and FSH, LH, and estradiol according to pubertal stage
Inhibin A Inhibin B FSH
a

Inhibin B LH
a

Estradiol
a

FSH

LH

Estradiol

Stage Stage Stage Stage Stage

I II III IV V

0.65 (0.07) 0.4 (0.05) 0.07 (0.79) 0.27 (0.02)

0.31 (0.30) 0.21 (0.23) 0.05 (0.78) 0.46 (0.0001)

0.58 (0.06) 0.71 (0.0002) 0.18 (0.34) 0.01 (0.87)

0.85 (0.07) 0.79 (0.0001) 0.76 (0.0001) 0.78 (0.0001)

0.27 (0.003) 0.29 (0.06) 0.11 (0.46) 0.77 (0.0001) 0.77 (0.0001)

0.59 (0.0002) 0.45 (0.005) 0.22 (0.12) 0.5 (0.001) 0.42 (0.0001)

0.41 (0.01) 0.73 (0.0001) 0.37 (0.06) 0.21 (0.22) 0.02 (0.79)

Correlations are given as Pearson correlation coefficients (P values) and are corrected for age and left-censoring (see Subjects and Methods, Statistics). Correlations that are significant at the 0.05 level are in italic. a Inhibin A was below the detection limits in 99% of subjects in stage I, so correlations are not calculated. TABLE 5. Specificities and sensitivities for determining whether a girl is in puberty (see Subjects and Methods) for different cut-off values of LH, estradiol, inhibin A, and inhibin B
Hormone and cut-off level Specificity (%) Sensitivity (%), stage II Sensitivity (%), stages II and III

LH 0.45 U/L LH 0.1 U/L LH 0.08 U/L Inhibin B 70 pg/mL Inhibin B 50 pg/mL Estradiol 40 pmol/L LH 0.45 U/L or inhibin B 70 pg/mL LH 0.1 U/L or inhibin B 50 pg/mL LH 0.08 U/L or inhibin B 40 pg/mL LH 0.08 U/L or estradiol 40 pmol/L

96 85 80.6 94.4 77.4 88.6 92.7 69.4 60 74

42.5 60 67.5 40 50 50 52.5 72.5 95 77.5

66.7 76.8 81.2 50.7 65 69.1 72.5 84 97 87

mL, respectively. In the menstrual cycle group all women had ovulated and had a functioning corpus luteum, whereas in the population of adult women some of the cycles may have been anovulatory, or the luteal phase may have been insufficient. Levels during the luteal-follicular transition (Fig. 4), regardless of whether the menstrual cycle was aligned to the day of the intercycle FSH peak or the first day of menses, showed changes in inhibin B that followed changes in FSH with a lag of 12 days. Inhibin B concentrations had risen considerably on the day of the FSH peak and peaked the day after. Estradiol levels increased more slowly, although an increasing tendency was evident from the day of the FSH peak, and levels had doubled 3 days after the FSH peak compared with levels 2 days before the peak. Estradiol levels were still increasing when both inhibin B levels and FSH levels were falling.

FIG. 3. Serum values of inhibin A, inhibin B, FSH, and LH throughout the menstrual cycle in 10 healthy women. Values are the mean and SEM. The days are aligned to the day of the LH peak (day 0).

Discussion

This cross-sectional study of 611 girls and women, aged 6 32 yr, shows the developments in levels of serum inhibin

SERUM INHIBIN A AND B IN GIRLS AND WOMEN

1639

FIG. 4. Serum values of inhibin B, FSH, and estradiol in 10 healthy women with normal menstrual cycles. Values are the mean and SEM. The days are aligned to the day of the intercycle FSH peak (day 0).

A and inhibin B as ovarian activity changes with the progression of puberty and establishment of the menstrual cycle. In our population, serum inhibin B was low or undetectable in Tanner breast stage I; however, finding measurable levels of inhibin B in over 60% of prepubertal girls indicates that some gonadotropin-responsive follicular activity is occurring at this stage. Moreover, inhibin B levels increased within Tanner stage I as a function of age, which could mean that activity increases as the girls come closer to showing clinical signs of puberty. Measurable estradiol levels may be found in 50% of the same population, but estradiol is also produced at other sites within the body (e.g. peripheral conversion of adrenal androgens), whereas inhibin B is specific for the ovary. Levels of inhibin B increase through pubertal stages II and III, corresponding to observations of inhibin B principally being produced by small antral follicles in response to gonadotropin stimulation (20). This could indicate more follicles being recruited as puberty progresses, follicles reaching a later stage of development before undergoing atresia, or both. All girls in Tanner stage III had measurable levels of inhibin B, and the median level was higher than that in adult women. Thus, Tanner stage III seems to represent a period of consistently high ovarian follicular activity before the development of the adult menstrual cycle with ovulation and a luteal phase. Serum inhibin A was undetectable in all girls staged as Tanner breast stage I with one exception; this girl also had

the highest stage I levels of inhibin B (180 pg/mL), estradiol (176 pmol/L), FSH (5.04 U/L), and LH (1.5 U/L), indicating that the hormonal changes of puberty had begun. Levels of inhibin A increased progressively from Tanner stage I into adulthood; however, levels corresponding to those seen in the midluteal phase in adult women (19 pg/mL) were only seen in girls that had experienced menarche. The decrease in inhibin B levels and the increase in inhibin A levels seen from Tanner stage III into adulthood could be explained by an increasing proportion of the subjects at each stage and age having regular, ovulatory menstrual cycles. This would give a progressively greater likelihood of being in the luteal phase on the day of blood sampling and thus having a high serum inhibin A and a low serum inhibin B. Manasco et al. (8), when measuring total inhibin, report a tendency for total inhibin levels to drop in Tanner IV before increasing again in Tanner V. Depending on the specificities of their total inhibin assay for inhibin A and inhibin B, this could be explained by our results as being the consequence of summing the falling tendency of inhibin B in stages IV and V with the rising tendency of inhibin A. Serum levels of inhibin B were considerably (medians less than half the value) lower in girls than boys at all stages of puberty using our reference values for boys in puberty (5), in agreement with previous findings (7, 8, 21). For clinical purposes, a basal hormone assay that could satisfactorily distinguish a pre-pubertal from a pubertal girl could be very useful. Unfortunately, like the other reproductive hormones, the overlap between levels of inhibin A and inhibin B in different Tanner stages is so great that measurement of inhibin A or inhibin B alone cannot fulfil this role. The sensitivities and specificities regarding whether a girl is in Tanner stage II or Tanner stage II/III for single measurements of inhibin B are no better than those for estradiol. Combining cut-off values for inhibin B and LH improves sensitivity, but at the cost of specificity, i.e. very few false negatives, but many false positives. Levels of inhibin A of 7 pg/mL (the detection limit) or more make it highly likely that the girl has entered puberty (high specificity); however, levels below 7 pg/mL can in no way exclude puberty (low sensitivity). Examining the changes in correlations between reproductive hormones through the stages of puberty may shed some light on the maturational processes of puberty. In early puberty, the reproductive hormones are all positively correlated to one another, reflecting the fact that all of the levels increase as puberty progresses. However, interestingly, LH is more strongly correlated with both inhibin A and inhibin B in early puberty than is FSH, supporting observations that increasing secretion of LH is an early and significant event in the onset of puberty. With respect to inhibin A, changes in mid- and late puberty include a negative correlation appearing between inhibin A and inhibin B and between inhibin A and FSH, reflecting that inhibin A is secreted out of phase with inhibin B and FSH within the menstrual cycle. With respect to inhibin B, the positive gonadotropin correlations of early puberty disappear in stage III to reappear in stage IV, now more strongly positively correlated to FSH than to LH, in agreement with FSH and inhibin B being

1640

SEHESTED ET AL.

JCE & M 2000 Vol 85 No 4

secreted in phase with one another within the menstrual cycle. We found a specific pattern of inhibin A and inhibin B secretion through the menstrual cycle. Inhibin B is principally secreted in the follicular phase, and inhibin A is principally secreted in the luteal phase; levels fall synchronously with those of progesterone, as previously reported (13, 14, 18, 19). There are some minor discrepancies in timing between our present results and the findings of Groome et al. (14). They report the postovulatory inhibin B peak to occur 2 days after the LH peak, whereas in our data it occurs the day after. This inhibin B peak is probably derived from follicular fluid released during ovulation. With respect to the timing during the luteal-follicular transition, Groome et al. (14) describe inhibin B levels as being unchanged on the day of the FSH peak, but rising thereafter to peak 4 days later. The study by Welt et al. (18) aligns the data to the first day of menses and shows inhibin B concentrations to increase more or less coincidentally with FSH. We found inhibin B concentrations to follow FSH concentrations with a lag of 12 days, and inhibin B levels peaked the day after FSH. The fact that inhibin B levels rise before estradiol levels during this period has been taken as suggesting that inhibin B could have a distinct role in modulating the intercycle FSH peak (14, 18). We also found that inhibin B levels peak well before estradiol levels. However, it is worth pointing out that estradiol is a very potent agent, and that levels of estradiol are clearly on the increase from the day of the FSH peak and have doubled by day 3 after the FSH peak. It is an open question as to how great an increase in estradiol is necessary to exert a suppressor influence on FSH secretion. In conclusion, inhibin A and inhibin B are gonadal hormones that are produced in a specific pattern in response to gonadotropin stimulation in females. Their role during pubertal development in girls is not known; they may have paracrine actions involved in the growth and maturation of the ovary as well as endocrine actions in the maturation of the hypothalamic-pituitary-gonadal axis. The pattern of inhibin secretion seen in girls during pubertal development is consistent with our knowledge of the pattern in adult normally menstruating women and reflects the average degree of ovarian follicular maturation at various stages of puberty. In physiological terms, inhibin B in prepubertal and early pubertal girls can be seen as a marker of gonadotropinstimulated early follicular activity. The appearance of measurable inhibin A can be seen as a marker for a follicle having at least matured to a stage corresponding to the late follicular stage in adult women. Although measuring serum inhibin A and inhibin B did not offer superior prediction of pubertal onset compared to LH and estradiol, they do give unique information about the level of follicular maturation on the day of the blood test and could thus contribute to the evaluation of puberty in a clinical setting, particularly in concert

with ovarian ultrasound counting the number and size of visible follicles.
References
1. McCullagh DR. 1932 Dual endocrine activity of the testes. Science. 76:19 20. 2. Robertson DM, Sullivan J, Watson M, Cahir N. 1995 Inhibin forms in human plasma. J Endocrinol. 144:261269. 3. De Kretser DM, Robertson DM. 1989 The isolation and physiology of inhibin and related proteins. Biol Reprod. 40:33 47. 4. Andersson A-M, Toppari J, Haavisto A-M, Petersen JH, Simell T, Simell O, Skakkebk NE. 1998 Longitudinal reproductive hormone profiles in infants: peak of inhibin B levels in infant boys exceeds levels in adult men. J Clin Endocrinol Metab. 83:675 681. 5. Andersson A-M, Juul A, Petersen JH, Mu ller J, Groome NP, Skakkebk NE. 1997 Serum inhibin B in healthy pubertal and adolescent boys: relation to age, stage of puberty, and follicle-stimulating hormone, luteinizing hormone, testosterone and estradiol levels. J Clin Endocrinol Metab. 82:3976 39782. 6. Illingworth PJ, Groome NP, Bryd W, Rainey WE, McNeilly A, Mather JP, Bremner WJ. 1996 Inhibin-B: a likely candidate for the physiologically important form of inhibin in man. J Clin Endocrinol Metab. 81:13211325. 7. Burger HG, McLachlan RI, Bangah M, et al. 1988 Serum inhibin concentrations rise throughout normal male and female puberty. J Clin Endocrinol Metab. 67:689 694. 8. Manasco PK, Umbach DM, Muly SM, Godwin DC, Negro Vilar A, Culler MD, Underwood LE. 1997 Ontogeny of gonadotrophin and inhibin secretion in normal girls through puberty based on overnight serial sampling and a comparison with normal boys. Hum Reprod. 12:2108 2114. 9. Tanner JM, Whitehouse RH. 1976 Clinical longitudinal standards for height, weight, height velocity, weight velocity, and stages of puberty. Arch Dis Child. 51:170 179. 10. Juul A, Bang P, Hertel NT, et al. 1994 Serum insulin-like growth factor-1 in 1030 healthy children, adolescents and adults: relation to age, sex, stage of puberty, testicular size and body mass index. J Clin Endocrinol Metab. 78:744 52. 11. Jensen TK, Hjollund NHI, Henriksen TB, et al. 1998 Does moderate alcohol consumption affect fertility? Follow up study among couples planning first pregnancy. Br Med J. 317:505510. 12. Juul A, Scheike T, Pedersen AT, Main KM, Andersson A-M, Pedersen LM, Skakkebk NE. 1997 Changes in serum concentrations of growth hormone, insulin, insulin-like growth factor and insulin-like growth factor-binding proteins 1 and 3 and urinary growth hormone excretion during the menstrual cycle. Hum Reprod. 12:21232128. 13. Groome NP, Illingworth PJ, OBrien M, Cooke I, Ganesan TS, Baird DT, McNeilly AS. 1994 Detection of dimeric inhibin throughout the human menstrual cycle by two-site enzyme immunoassay. Clin Endocrinol (Oxf). 40:717723. 14. Groome NP, Illingworth PJ, OBrien M, Pai R, Rodger FE, Mather JP, McNeilly AS. 1996 Measurement of dimeric inhibin B throughout the human menstrual cycle. J Clin Endocrinol Metab. 81:14011405. 15. Yu K, Jones MC. 1998 Local linear quantile regression. J Am Stat Assoc. 93:228 237. 16. Tobin J. 1958 Estimation of relationships for limited dependent variables. Econometrika. 26:24 36. 17. Cavallo A, Zhou XH. 1994 LHRH test in the assessment of puberty in normal children. Horm Res. 41:10 15. 18. Welt CK, Martin KA, Taylor AE, et al. 1997 Frequency modulation of folliclestimulating hormone (FSH) during the luteal-follicular transition: evidence for FSH control of inhibin B in normal women. J Clin Endocrinol Metab. 82:26452652. 19. Welt CK, McNicholl DJ, Taylor AE, Hall JE. 1999 Female reproductive aging is marked by decreased secretion of dimeric inhibin. J Clin Endocrinol Metab. 84:105111. 20. Roberts VJ, Barth S, El-Roeiy A, Yen SSC. 1993 Expression of inhibin/activin subunits and follistatin messenger ribonucleic acids and proteins in ovarian follicles and the corpus luteum during the human menstrual cycle. J Clin Endocrinol Metab. 77:14021410. 21. Crofton PM, Illingworth PJ, Groome NP, et al. 1997 Changes in dimeric inhibin A and B during normal early puberty in boys and girls. Clin Endocrinol (Oxf). 46:109 114.