| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Obstetrics, Gynecology, and Women’s Health, Albert Einstein College of Medicine, Bronx, New York 10461
Address all correspondence and requests for reprints to: Nanette Santoro, M.D., Division of Reproductive Endocrinology, Department of Obstetrics, Gynecology, and Women’s Health, 1300 Morris Park Avenue, Mazer 316, Bronx, New York 10461. E-mail: glicktoro{at}aol.com.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Methods: Daily urine and menstrual records were collected for 2 yr each from 10 girls aged 11–13 yr at study entry. Urinary excretion of LH, FSH, estradiol (E1c), and progesterone (Pdg) metabolites was measured using established ELISAs. An objective algorithm detected rises of LH, FSH, E1c, and Pdg consistent with follicular maturation and/or ovulation.
Results: Nine of 10 girls enrolled into the study experienced the onset of menarche prior to or during the 2-yr collection period. LH and FSH surges, as well as small amplitude Pdg increments, were observed prior to menarche. Regular, ovulatory-appearing cycles with LH surges and gradually increasing and more sustained Pdg rises were observed over time after menarche, although duration of Pdg elevations remained shorter than in adult women (8.9 ± 1.0 vs. 12.1 ± 0.8 d, P = 0.043). E1c levels leading to LH/FSH surges were lower in perimenarcheal girls than adult controls, and bleeding episodes did not uniformly correlate with hormone patterns. Progressive increases in FSH and Pdg, but not LH or E1c, were observed in association with menarche.
Conclusion: Mature hormone patterns are established within several months of and even prior to menarche in normal-weight perimenarcheal girls. Factors determining menstrual bleeding in perimenarcheal girls may not be solely dependent on reproductive hormones or the neuroendocrine axis.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
One study that reviewed menstrual records of 250 girls, 100 of whom were studied longitudinally for up to 6 yr (3140 menstrual intervals), indicated that menstrual regularity was established within 2 yr after menarche in the majority (6). Others suggest that up to 5–7 yr of gynecological age are required to establish a reliable ovulatory pattern (7, 8, 9). More recent studies using questionnaires and hormone assessments suggest that regular menstruation is established within 6–12 months of menarche (10, 11). Although menstrual irregularity in adolescent girls is often attributed to an immature H-P-O axis, many adolescents with menstrual abnormalities have an increased risk of continued menstrual abnormality and reproductive disorders (12, 13, 14).
Our working hypothesis is that H-P-O axis maturation is an early event in nutritionally replete, normal-weight adolescent women. To begin to test this hypothesis, we performed a detailed longitudinal pilot study of daily urinary hormone excretion in perimenarcheal girls.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Ten healthy volunteers were recruited from the community, using university-wide advertising and oral communications and met the following criteria: 1) age 11–13 yr at study entry, 2) premenarcheal or menarche within 3 months before enrollment, 3) Tanner stage 3 or greater (both breast and pubic hair), 4) height within the fifth to 95th centile of normal for age, 5) body mass index (BMI) for age between the fifth and 95th percentile (15), 6) no personal or family history of hyperandrogenism or polycystic ovary syndrome, and 7) normal prolactin and TSH.
Volunteers meeting these criteria were interviewed by an experienced researcher. Participants were excluded for the following reasons: 1) exercise 4 h/wk or more; 2) one or more first-degree relatives with type 2 diabetes or polycystic ovary syndrome (defined as irregular menstrual cycling and hirsutism or hyperandrogenism); 3) acanthosis nigricans; 4) obesity, defined as BMI for age greater than the 95th percentile [except for participant 2, who reported a BMI of 23 kg/m2 (85%tile) at screening but had a BMI of 28 kg/m2 at examination (>95th percentile, BMI 28 kg/m2)]; 5) history of renal or hepatic disease; and 6) recent voluntary weight loss of 1 lb/week or more. The final sample consisted of six Caucasian, three Hispanic, and one African-American perimenarcheal girls. The study was approved by the institutional review board; all participants and parents or guardians gave informed consent before entry.
Adult controls
Fourteen healthy, regularly cycling women aged 18–38 yr were used as a comparison group. Eleven have been previously reported (16). An additional three control women were added in 2006 along with other quality control steps to assure that the assay has not changed substantially over the past 14 yr. Their results were identical with the original 11. All adult controls had monthly menses at intervals of 25–35 d; did not meet exclusion criteria 1, 5, and 6 above; had normal screening prolactin and TSH; and were free of hirsutism, virilization, or diabetes.
Urine specimen and menstrual calendars collection
Participants collected daily urine samples for a total of 2 yr. These techniques have been reported previously and have been extensively validated (16, 17). Briefly, polypropylene tubes were used with glycerol as a preservative to yield a final concentration of 7% when tubes were filled to a premarked line. Participants were supplied with 90–120 tubes at a time, storage boxes, and written instructions. Each participant collected first-morning voided urine, transferred it to the tube, and froze it within 2 h. Errors or irregularities in collection were recorded on the menstrual calendars.
Participants were asked to record menstrual flow and return monthly calendars. Flow was categorized as either spotting or bleeding on a supplied calendar. Monthly contacts were made to check for any problems or whether menarche had occurred. Specimens and calendars were picked up at least every 3 months. To continue in the study, participants had to provide at least 80% of all possible samples.
Urinary excretion of LH, FSH, E1c metabolites (a mixture of estrone sulfate and estrone glucuronide) and progesterone metabolites (Pdg, pregnanediol) were measured in duplicate using previously established ELISAs using antisera and conjugate tracers provided by Dr. Bill Lasley (University of California, Davis) (18). LH and FSH were measured using a solid-phase, two-site fluoroimmunometric assay (DELFIA; Pharmacia, Gaithersburg, MD) adapted for urine specimens as previously described (19). Urine was assayed neat for gonadotropins and diluted 1:10–11:100 for steroid assays to assure that samples read at or below the ED50 on the standard curve, standardized to creatinine (20), and adjusted for the added glycerol. Intra- and interassay coefficients of variation were: LH: 7.7 and 18.7%; FSH: 9.7 and 16.2%; E1c: 10.3 and 7.9%; and Pdg: 8.2 and 14.5%, respectively.
Data analysis
Data were organized into Excel spreadsheets (Microsoft, Redmond, WA) and inspected for error and completeness. Previously developed algorithms that combed the hormone series for significant rises in LH, FSH, E1c, and Pdg consistent with midcycle surges, follicular maturation and ovulation, respectively, were adapted for use in these samples. Algorithms previously used in SAS (Cary, NC) (21, 22) were rendered functional on SPSS software (version 13.0; SPSS Inc., Chicago, IL). Briefly, to determine ovulation, the data are combed to define the lowest 5-d average of Pdg. An ovulatory rise is defined as a 3-fold increment above this baseline nadir mean that is sustained for at least 3 d (22). Previous validation of this method in midreproductive aged and perimenopausal women confirmed its robustness (21). For LH, FSH, and E1c rises, a 5-d moving average is similarly computed and a rise of at least 3 SD above the moving average for at least 2 d is defined as a surge (for LH and FSH) and a rise (for E1c).
Because these algorithms have not been used before to evaluate hormonal dynamics in pubertal girls, we performed some systematic variations in detection criteria for significant hormone rises to check for robustness of the thresholds. We varied the number of SDs or the number of multiplications of the Pdg nadir (from 2 to 4) required to define a hormone rise or surge. We also varied the number of days (from 4 to 6) of the moving average to which the surge or rise values were compared. A 5-d moving average and a 3 SD threshold was reconfirmed as adequate for detecting surges of LH and FSH and rises in E1c. For Pdg rises, applying the Kassam (22) algorithm without regard to its amplitude resulted in the detection of many rises in Pdg in premenarcheal girls that were of very low amplitude [less than 0.5 µg/mg creatinine (Cr)]. For this reason, we constrained Pdg rises to a minimum amplitude of 1.0 µg/mg Cr. This best represented the visual data and did not change the detection of Pdg rises in the adult controls.
Missing data, problematic for all of the reported detection algorithms, were handled in several different ways. If the missing data resulted from an undetectable assay (11 in participant 3, 16 in participant 4, one in participants 5 and 9, and four in participant 10), it was read by the algorithm as missing and assigned a value of zero (no significant rise). In calculating the moving average, this missing point was disregarded by the algorithm. If more than two consecutive points were missing in calculating a moving average, spurious surges resulted when the mean values for the series were relatively high, compared with the limit of detection; these spurious surges were edited out of the data before analysis. If missing data resulted from samples that were never collected, the series was manually adjusted to make the data points contiguous. Occasionally, clear-cut elevations in Pdg consistent with ovulation were observed immediately after a period of missing values. In these instances, there was no valid prior baseline from which a moving average could be calculated; to correct for the missing data, the hormonal nadir in the data series before the next detectable rise was assigned manually as the baseline to allow for a subject-specific comparison. Then, either a 3-fold or a 1.0 µg/mg or greater Cr rise from the manual baseline was considered to be positive for luteal activity because no SD could be calculated from the interpolated nadir. When the data were reanalyzed assigning undetectable values the same level as the detection limit of the assay, identical results were obtained.
Descriptive characteristics of cycle parameters are expressed as means and interquartile range, SD, or SEM, as indicated. Summary hormonal cycle statistics were compared with adult controls using Wilcoxon rank sum tests. Group mean comparisons were performed using t tests or Wilcoxon testing when data were not normally distributed. Data were standardized to time of menarche for chronological analyses. For these data, repeated measures one-way ANOVA or Kruskal-Wallis testing was used. SPSS 13.0 software and STATA (version 9; STATA Corp., College Station, TX) were used for all analyses. P < 0.05 was considered statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
. The reported BMI at telephone screening, used as the entry criterion, differed from the measured BMI in three cases. One girl whose BMI was greater than the 95th centile did not have her data reported with the rest of the group (see below). The other had a BMI of 29 kg/m2, slightly above the 95th centile and one girl had a measured BMI of 14.6 kg/m2, slightly below the fifth centile. Ten girls were recruited. Two had menarche within 3 months before study enrollment; one provided urine for only 3 months, and the other did not return her menstrual calendars for the first 6 months. One girl sustained a head injury 1 month after beginning collection and then developed oligomenorrhea and was subsequently diagnosed with polycystic ovary syndrome. Her data were excluded from analysis. One girl dropped out of the study within 2 months of study enrollment and never initiated urine collection. Of the six remaining girls, all of whom collected urine sample for the full 2 yr, one did not commence menstruation during the collection. Therefore, six of seven possible menstrual calendar series that included or were subsequent to menarche were available for analysis of intermenstrual intervals. Complete data, i.e., full 2-yr urine collections plus informative menstrual calendar data, exist for six girls. Complete hormone data exist for seven girls, and partial hormone data exist for one. Of 4176 total possible samples that could be collected, 725 were missing, for overall compliance of 82.64% (range 76.6–92.2%).
|
Hormonal patterns before and after menarche
Figure 1 shows mean LH, FSH, E1c, and Pdg in 3-month intervals relative to the onset of menarche in eight girls. Although both FSH (P = 0.29) and Pdg (P = 0.33) increased over time, neither was statistically significant. Basal Pdg in premenarcheal girls was 0.71 ± 0.47 (SD), 0.96 ± 0.43 (SD) in postmenarcheal girls and 1.37 ± 0.99 (SD) in adult controls (P = 0.087 for comparison between premenarche and adults). Although means did not differ, the within-participant variance in day-to-day Pdg was significantly greater in postmenarcheal girls, compared with premenarcheal girls (P = 0.028). A trend for an increase in LH, E1 g and Pdg is observed with progressive postmenarcheal cycles (Fig. 2).
|
|
Pdg peak excretion characteristics for each postmenarcheal participant are summarized in Table 2. Small amplitude Pdg increments were observed before menarche (mean 1.33 ± 0.10 µg/mg Cr, range 1.01–1.98 µg/mg Cr), which increased to a mean of 3.88 ± 0.87 (SEM) µg/mg Cr (range 1.07–8.98 µg/mg Cr) within the first 6 months after the onset of menarche. These increments increased slightly 6–12 months after menarche to a mean of 3.95 ± 0.73 (SEM) µg/mg Cr (range 1.85–6.62 µg/mg Cr), and all were lower than adult controls (mean 22.18 ± 5.75 µg/mg Cr, range 5.1–69.3) at all time points (P < 0.000).
|
LH surges were observed before menarche in five girls. Premenarcheal LH surges [21.28 ± 3.90 (SEM) mU/mg Cr] did not differ from adult controls, but postmenarcheal girls had higher surge amplitude than premenarcheal girls [40.16 ± 4.97 (SEM) mU/mg Cr; P = 0.01]. One girl (participant 2; Fig. 3) consistently generated LH surges at a magnitude (70.3–102.4 mU/mg Cr) comparable with adult controls within 6 months after menarche. Despite this relatively early appearance of adult-pattern LH surges, no correlation was observed between the magnitude of LH surges and either the magnitude or duration of the subsequent Pdg rise.
|
Baseline E1c excretion before menarche was 11.34 ± 2.66 (SEM) ng/mg Cr in all girls, compared with the early follicular phase mean of 15.02 ± 1.85 (SEM) ng/mg Cr in adult controls (P = 0.363). After menarche, mean E1c increased to 20.05 ± 2.12 (SEM) ng/mg Cr within 3 months and increased slightly to 21.11 ± 4.12 (SEM) ng/mg Cr by 6 months after menarche but remained significantly lower than mean E1c excretion in adult controls [39.71 ± 2.92 (SEM) ng/mg Cr, P = 0.001].
Relationship of hormone patterns to menstrual episodes
Table 2
indicates the types of patterns that were observed in each of the observed menstruations. Nine of 26 cycles (34.6%) conformed to an adult-type pattern and were preceded by an E1c rise and an LH surge and followed by a Pdg rise (Fig. 4). The remaining menstruations had a variety of combinations of these features. Cycles with an LH surge and algorithm-detected E1c but no Pdg rise (Fig. 5) and cycles with no detectable change in any of the three hormones (Fig. 6) were observed in 11.5% of menstruations recorded.
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Although menstrual calendars and basal body temperature charts are clinically helpful for establishing the presence of menstrual cycling and ovulation, neither provides sufficient information on the interactions of the three components of the H-P-O axis at the hormonal level. Urinary measurements are useful, noninvasive alternatives to serial blood sampling for assessing hormonal status over long periods of time (16, 17, 18, 21, 23).
Prior epidemiological studies have reported that regular menstruation takes from as few as 2 (10) to as many as 7 yr after menarche (6, 7, 8). Many such studies involved girls who differ from the current U.S. population. Prior studies included few minority women, some of whom attain menarche and undergo pubertal maturation earlier than Caucasians, in part related to increased BMI (24). Weight and BMI were not recorded for some prior studies, some of which involved institutionalized girls (6, 7, 12, 14). It may be that a relatively large proportion of girls in some of these studies (6) were nutritionally marginal, factors that may have led to the findings of relatively slow establishment of regular, ovulatory menstrual cycle patterns. Our sample included girls who were nutritionally adequate to obese, reflective of the current U.S. population (25). The girl with the lowest baseline BMI in our study was the last in the cohort to experience menarche. To the extent that the increase in body fat in adolescent women in the United States over the past century (26) can be viewed as a permissive signal to the hypothalamic-pituitary axis, elimination of undernutrition might be expected to reduce the overall age at menarche. Some but not all (23) studies of age at menarche support this notion.
Our findings are most comparable with a larger study of 112 Caucasian perimenarcheal girls in Pennsylvania, who were followed up every 6 months (10). Maturation of the reproductive axis, as defined by a single urinary progesterone metabolite sample consistent with adult levels, was reportedly achieved within 1 yr after menarche in the majority of the girls. A 6-month frequency of sampling is insufficient to detect monthly ovulations. Our findings of low perimenarcheal Pdg excretion relative to adult women might also have caused an underdetection of ovulation in the Pennsylvania sample. Our results are also consistent with previous findings by Metcalf et al. (9) and Borsos et al. (27), who observed a mature pattern of ovulatory cycles in 23 and 13% women within 1 yr after menarche, respectively. However, these studies are also limited by the use of only weekly urine sampling, which may be inadequate for the detection of some of the brief Pdg rises we observed. These authors considered only luteal phases of greater than 7 d duration to be ovulatory. These more recent cohort studies are in overall agreement with ours in indicating that ovulatory competency is established more rapidly than is widely appreciated. The even more rapid transition to ovulatory competency we observed is likely due to our more intensive sampling paradigm. We were able to demonstrate a progressive increase in the duration of luteal phase, from 2 to 4 d to a length compatible to that of the adult controls (11–12 d), over several months after initial menstruation and an increase in overall Pdg output per luteal phase, indicative of increasing luteal sufficiency over time.
Two girls generated LH surges of an adult magnitude of 26.7 and 21.8 mU/mg Cr with estrogen and progesterone rises, before the onset of menarche. Two other girls also produced LH peaks in response to rises in E1c without a subsequent Pdg rise. These peaks ranged from 21.8 to 47.4 mIU/mg Cr, also consistent with adult, preovulatory LH surges. This discovery would not have been possible without using daily sampling because weekly or semiannual sampling would be highly unlikely to capture events of such short duration. It also indicates that the permissive signals required for acquisition of a positive feedback response are present before hormone patterns are organized into a normal menstrual cycle. Postmenarcheal girls in our study mounted normal-appearing LH surges.
Normal-appearing LH surges and estrogen elevation occurred without significant progesterone metabolite rises in two girls. This finding indicates a possible dyssynchrony between maturation of the H-P-O axis and full ovarian competency and is congruent with observations that irregular menstruation is associated with gynecological age, low BMI, chronic nonspecific lung disease or allergic disease, weight loss, and stress (28). Although most of these factors were excluded in our initial screen, psychosocial stress and allergic diseases were not specifically sought and might have influenced menstrual bleeding patterns. Our observation of small episodic increments in Pdg excretion before menarche supports the working hypothesis of Rosenfield and Barnes (5), that low-amplitude hormonal cycles may occur at early puberty with nighttime levels greater than daytime, although no monthly cyclicity was observed in our participants. These premenarcheal Pdg rises may reflect the normal process of ovarian maturation, which may be an amplification process, in that all the hormonal components are functioning at an adult frequency, but the hormonal signals are dampened and slowly increase to detectable levels.
We also observed menstrual bleeding in girls with low E1c and no evidence of Pdg withdrawal in one instance. This finding suggests that factors determining menstrual bleeding in perimenarcheal girls may not be exclusively dependent on ovarian hormones or may occur in the presence of very small and/or chronic changes in concentrations of E1c and/or progesterone that are undetectable in our urinary sampling paradigm.
The early postmenarcheal cycles we describe herein resemble the infertility syndrome of luteal phase defect (29). However, the onset of normal-appearing LH surges in close temporal association to estrogen elevations before these short Pdg rises in our participants strongly supports the concept that the hypothalamic-pituitary axis is appropriately reactive to the process of folliculogenesis. Whereas a rise in progesterone signifies the development of the corpus luteum after ovulation, a short luteal phase observed after a normal appearing estrogen rise and gonadotropin surge suggests the possibility that the ovary fails to develop an adequate corpus luteum despite normal folliculogenesis. In other words, the neuroendocrine axis may well be comparably matured and similar to adult women within 6 months of menarche, and relative ovarian immaturity accounts for intermittent ovulation. It is also possible that the short Pdg rises we observed close to menarche reflect suboptimal LH surges. However, this seems less likely because the LH dynamics we observed in these girls are well within the normal range (30).
Girls at both extremes of BMI were more likely to have later onset of menarche. This result is partially consistent with previous findings (10) as well as the notion that body weight or body fat is a permissive maturational signal to the hypothalamic-pituitary axis. Our sample size is too small to draw a definitive conclusion, but the possibility that girls with a high BMI are as of yet undiagnosed with ovulatory disorders is supported by our observation that the girl with the highest BMI was ultimately diagnosed with polycystic ovary syndrome.
Our study has several limitations. We were unable to observe directly ovulation events using a definitive technique such as ultrasonography. This would have helped clarify whether the hypothalamus-pituitary and the ovaries mature sequentially, simultaneously, or independently. The sample size also limits generalizability of these findings, which require confirmation in larger samples of young perimenarcheal women. Despite appealing directly to the mothers of the participants to help assure compliance with the protocol, and routine, frequent phone contacts, there is still a small amount of missing data, especially during the summer months when many of the girls went on family vacations. We also experienced difficulty maintaining the menstrual records of two of the participants (lost in the mail). We mitigated this problem by subsequently instructing all participants to save their menstrual calendars for pick-up with urine specimens.
In summary, we have shown that regular ovulatory menstruation patterns can be established within several months of menarche. A time lag may exist between relatively early maturation of the hypothalamic-pituitary and the ovarian response to LH. Factors in addition to reproductive hormones may play a role in determining the onset of menstrual bleeding. Our findings bring new understanding into the physiological events of the maturation of the female reproductive system.
|
Footnotes |
|---|
Disclosure Statement: The authors have nothing to disclose.
irst Published Online February 5, 2008
Abbreviations: BMI, Body mass index; Cr, creatinine; E1c, estradiol; H-P-O, hypothalamic-pituitary-ovarian; Pdg, progesterone.
Received August 17, 2007.
Accepted January 24, 2008.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
