Predictive value of menstrual cycle pattern, body mass index, hormone levels and polycystic ovaries at age 15 years for oligo-amenorrhoea at age 18 years

M.H.A. van Hooff1,2,5, F.J. Voorhorst3, M.B.H. Kaptein1, R.A. Hirasing3, C. Koppenaal4 and J. Schoemaker1

1 Research Institute for Endocrinology, Reproduction and Metabolism, Division of Reproductive Endocrinology and Fertility, Vrije Universiteit Medical Center, Amsterdam, De Boelelaan 1117, 1081 HV Amsterdam, 2 Sint Franciscus Hospital, Rotterdam, 3 Institute for Research in Extramural Medicine, Vrije Universiteit Medical Center, Amsterdam and 4 Department of Youth Health Care of the Public Health Care Service, Amstelland-de Meerlanden, Amstelveen, The Netherlands

5 To whom correspondence should be addressed. e-mail: marcelvanhooff@planet.nl


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
BACKGROUND: On the question of how to counsel adolescents with irregular menstrual cycles or oligomenorrhoea, no clear answer has been given. Adolescents with oligomenorrhoea especially show endocrine abnormalities and may be at risk for ovulatory dysfunction and the polycystic ovary syndrome in adulthood. METHODS: We followed a cohort of adolescents to document changes in menstrual cycle pattern between ages 15 and 18 years in the general population. RESULTS: Two per cent (2/128) of adolescents with regular menstrual cycles developed oligomenorrhoea, and 12% (17/148) of those with irregular menstrual cycles did so. Fifty-one per cent (34/67) of the oligomenorrhoeic adolescents remained oligomenorrhoeic. Increase in body mass index (BMI), concentration of LH, androstenedione or testosterone, and polycystic ovaries (PCO) were associated with persistence of oligomenorrhoea. In multivariate analysis only a normal to high BMI (>19.6 kg/m2) consistently contributed significantly to predict persistent oligomenorrhoea. Glucose:insulin ratio as a marker for insulin resistance was not associated with an increased risk for oligomenorrhoea. CONCLUSIONS: Oligomenorrhoea at age 18 years is better predicted by menstrual cycle pattern at age 15 years than by LH or androgen concentrations or PCO at this age. Not only obese, but also normal weight oligomenorrhoeic, adolescents have a high risk of remaining oligomenorrhoeic.

Key words: adolescents/body mass index/menstrual cycle/oligomenorrhoea/polycystic ovaries


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Studies on variation in the length of the menstrual cycle suggest that complete maturation of the hypothalamic–pituitary–ovarian axis takes up to ~5 years after menarche (Treloar et al., 1967Go; Vollman, 1977Go). From this, it has been concluded that irregular menstrual cycles in the first 5 years after menarche are physiological, and no reason for clinical or endocrine evaluation (Gardner, 1983Go). However, about half the adolescents with oligomenorrhoea remained oligomenorrhoeic during a follow-up of 8 years (Southam and Richart, 1966Go). Furthermore, in the peri-menarcheal period the first symptoms of the polycystic ovary syndrome (PCOS) appear and in the first years after menarche the clinical picture may evolve or already be complete (Franks, 2001Go).

Adolescent girls with irregular menstrual cycles and particularly those with oligomenorrhoea have high LH and androgen levels and polycystic ovaries at ultrasound (Siegberg et al., 1986Go; Vihko and Apter, 1990Go; Venturoli et al., 1995Go; van Hooff et al., 1999bGo; Veldhuis et al., 2001Go). Hyperinsulinaemia has been documented in a small group of oligomenorrhoeic adolescents with extreme obesity (Apter et al., 1995) and in a group of lean an- or oligo-ovulatory hyperandrogenaemic adolescents (Ibanez et al., 2001Go).

The relationship between ‘physiological adolescent anovulation’ and ovulatory dysfunction due to PCOS remains to be defined (Rosenfield et al., 2000Go). To improve the knowledge about this, we estimated the predictive value of the menstrual cycle pattern, hormone concentrations and ultrasound pattern of the ovary at age 15 years for ovulatory dysfunction presenting as oligo-amenorrhoea at age 18 years in a stratified sample of the general population.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The POMP study (Pubertal Onset of Menstrual cycle abnormalities: a Prospective study) is an observational study on the natural course of pubertal onset menstrual cycle disorders. The study has a nested case–control design. Figure 1 gives a schematic summary of the methodology of the study that has been described in detail earlier (van Hooff et al., 1998aGo,b, 1999b).



View larger version (61K):
[in this window]
[in a new window]
 
Figure 1. Schematic presentation of the methodology.

 
Definitions of menstrual cycle patterns, the rationale for the classification of menstrual cycle patterns and the choice of oligo-amenorrhoea as primary endpoint
The definitions of the various menstrual cycle patterns were: regular menstrual cycles (RMC): an average length of the cycle between 22 and 41 days; either none or a single cycle with a length <22 or >41 days during the past year; irregular menstrual cycles (IMC): an average length of the cycle between 22 and 41 days; two or more cycles with a length <22 or >41 days during the past year; oligomenorrhoea: an average length of the cycle between 42 and 180 days; secondary amenorrhoea: the absence of menstruation for ≥180 days; polymenorrhoea: an average length of the cycle of ≤21 days. For every girl an estimate was made of the average length of the menstrual cycle.

In the analysis the IMC group was further divided into subgroups with an estimated average length of the menstrual cycle of 22–34 days and 35–41 days. Exact menstrual data were not always available. We coded adolescents with IMC and frequently cycles >6 weeks as an estimated average length between 35 and 41 days. For polymenorrhoea and oligomenorrhoea, definitions frequently used in the literature were chosen. An average length of the menstrual cycle between 22 and 41 days was defined as normal as the majority of cycles with lengths within this range are ovulatory (Vollman, 1977Go). We divided the group of girls with a normal average length on the basis of the number of menstrual cycles per year outside the normal range into the RMC and IMC group. This enabled us to group together those girls with the most RMC. This strategy resulted in optimal separation between the RMC group, the polymenorrhoea group and the oligomenorrhoea group. In addition, the IMC group consisted of young women in whom it was not clear whether we should classify the menstrual cycle pattern as normal or abnormal. This approach minimizes selection bias in the RMC group which is the reference category.

Oligomenorrhoea was chosen as the primary outcome measurement of the study as 90% of adult infertility patients with oligomenorrhoea have high concentrations of LH or androgens compatible with PCOS. Additionally, up to 90% of the oligomenorrhoeic subfertility population has polycystic ovaries at ultrasound (Franks, 1995Go).

We hypothesized that the risk for oligomenorrhoea at age 18 years depends on the menstrual cycle pattern at age 15 years and the presence of clinical, endocrine and ultrasound determinants of PCOS (van Hooff et al., 1998bGo).

Physical examination, hormone measurements and pelvic ultrasonography
Physical examination is described in detail elsewhere (van Hooff et al., 1999bGo). Abnormal body hair was defined as a Ferriman and Gallwey score of ≥1 and hirsutism as ≥8. Acne was scored according to Plewig and Kligman (1975Go). Acne was defined as a Plewig and Kligman score of ≥1, serious acne as a Plewig and Kligman score of ≥2.

In the first phase of the study, blood was taken at school between 12:00 and 17:00. In the second and third phases of the study, blood was taken after an overnight fast between 08:00 and 10:00 at school or at home. In girls with regular or irregular menstrual cycles, the blood was taken between the first and the 10th day of the menstrual cycle. To exclude the influence of a mid-cycle LH peak, blood from these girls should, in retrospect, have been taken ≥18 days before the next menstruation. The date of the period after venepuncture was verified by telephone calls. In oligomenorrhoeic girls, blood should have been taken ≥2 weeks after the first day of a period and ≥3 weeks before the next period. This procedure excludes the possible influence of peri-ovulatory hormonal changes and post-ovulatory progesterone production on the LH and androgen concentrations, which extends into the follicular phase of the next menstrual cycle (Minakami et al., 1988Go; Anttila, 1992Go; Taylor et al., 1997Go; Martins et al., 1998Go; van Hooff et al., 1999aGo). In secondary amenorrhoea, blood samples were not scheduled.

All hormones were determined by commercially available kits described in detail earlier (van Hooff et al., 1999bGo, 2000bGo). Glucose was measured by automatic hexokinase method. Insulin and glucose concentrations were only measured in samples taken after an overnight fast in the second and third phase of the study.

For a detailed description of the technique of pelvic ultrasonography and a description of the rationale for classification of the ultrasound patterns, see van Hooff et al. (2000aGo). Three girls with RMC, four with IMC and two with oligomenorrhoea showed one or more cysts of 15 mm and were excluded from the analysis as these cysts have a major influence on ovarian volume and may also show deviating hormonal activity and bias the results.

Follow-up
Table I shows the response to the follow-up questionnaires distributed after 18 and 36 months stratified by menstrual cycle pattern at the time of the first questionnaire. The mean age at the first, second and third phase of the study was 15.3 ± 0.6, 16.4 ± 0.7 and 18.1 ± 0.6 years respectively. The mean gynaecological age (calendar age minus age at menarche) was 2.0 ± 1.2, 3.3 ± 1.1 and 4.9 ± 1.1 years respectively.


View this table:
[in this window]
[in a new window]
 
Table I. Response to follow-up questionnaires on the menstrual cycle pattern mailed 1.5 years and 3 years after the initial questionnaire categorized by menstrual cycle pattern at the start of the study
 
Statistics
The data were analysed with BMDP statistical software package (BMDP statistical software, Ireland).

Baseline characteristics in adolescents with and without oligomenorrhoea at age 18 years categorized by menstrual cycle pattern at the start were compared by one-way analysis of variance (ANOVA) followed by post hoc t-tests.

We estimated the value of the menstrual cycle pattern at age 15 years and of various clinical, endocrine and ultrasound determinants to predict oligomenorrhoea at age 18 years by logistic regression analysis and proportional hazard analysis. In the proportional hazard analysis, time was expressed in gynaecological age. OC users were censored at the gynaecological age at which they started to use OC (see ‘Oral contraceptive use’).

In both logistic regression and proportional hazard analysis, we modelled body measurements and endocrine parameters at the time of the first questionnaire, and basal insulin concentrations, glucose:insulin ratio (GI ratio) and ultrasound pattern of the ovaries obtained at the time of the second questionnaire to predict oligomenorrhoea at the end of the follow-up. Not all data were available for all participants. The number of participants in each analysis is given in the tables.

When appropriate, body measurements, endocrine and ultrasound data were first modelled as continuous variables; subsequently the analysis was repeated after categorizing the data into two categories with threshold values at the 95th centile of the RMC cycle group and at the 50th centile of the menstrual cycle pattern group involved in the analysis.

P < 0.05 was considered to be statistically significant.

Oral contraceptive use
Our study was entirely observational. We explained that oral contraceptives (OC) use would interfere with the aim of the study, but the decision to start with OC was made by the girl, her parents and her general physician who had no information about the results of endocrine evaluation or ultrasonography.

The prevalence of OC use at the end of the follow-up was high (61%). The mean duration of OC use after 3 years follow-up was 1.3 ± 0.8 years. To explore the possibility of selection bias by menstrual cycle pattern in start of OC, {chi}2-tests were performed to compare the proportions of various menstrual cycle patterns in the year before the start of OC in OC users and the menstrual cycle pattern at the end of the study of girls with a complete follow-up. We found no significant differences. Thus there was no clear distinction by menstrual cycle pattern at the time adolescents started to use OC.

In logistic regression, OC users were recoded to the menstrual pattern in the year before OC use and included in the analysis to add to the power of the study. OC use was added as a variable in all multivariate analyses. In proportional hazard analysis, OC users were censored at the gynaecological age at which they had started OC use. This approach minimizes the chance that the conclusions are influenced by OC use.

Endocrine data of OC users are not in the analysis.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Table II shows changes in the menstrual cycle patterns after 3 years follow-up. Girls with IMC and an average length of the menstrual cycle between 35 and 41 days developed oligomenorrhoea in 34% (13/38) of cases. This is significantly more frequent than 4% (4/110) of the girls with IMC and an average cycle length between 22 and 34 days ({chi}2-test, P < 0.001). The frequency in the latter group is not significantly different from the RMC group (1.6%).


View this table:
[in this window]
[in a new window]
 
Table II. Menstrual cycle patterns after 3 years follow-up including oral contraceptive (OC) users who are recoded to the menstrual cycle pattern in the year preceding the start of OC
 
Table III shows the odds ratio (OR) and 95% confidence intervals (95% CI) of the prevalence of oligomenorrhoea at age 18 years for girls with IMC or oligomenorrhoea at the start of the study compared with those with RMC at the start of the study.


View this table:
[in this window]
[in a new window]
 
Table III. Odds ratios (95% confidence intervals) of the prevalence of oligomenorrhoea at age 18 years for girls with irregular menstrual cycles (IMC) or oligomenorrhoea at the start of the study compared with those with regular menstrual cycles (RMC) at the start of the study
 
In logistic regression analysis including all girls with RMC, IMC and oligomenorrhoea, menstrual cycle pattern at age 15 years was a better predictor for oligomenorrhoea at age 18 years [likelihood ratio test (LRT) = 45.6, df = 2, P < 0.0001; for odds ratios see Table III] than high testosterone concentration, i.e. a concentration >95th centile of the RMC group [LRT = 6.0, df = 1, P = 0.01; OR (95% CI) = 4.1 (1.3–12.7)], high LH concentration [LRT = 8.0, df = 1, P = 0.001; OR (95% CI) = 3.8 (1.5–9.2)], or polycystic ovaries at ultrasound [LRT = 22.8, df = 1, P < 0.001; OR (95% CI) = 6.2 (2.7–14.3)]; high androstenedione concentrations, hirsutism, acne or obesity showed no predictive value in this model.

As the risk for oligomenorrhoea at an age of 18 years was mainly determined by the menstrual cycle pattern at the start of the study, further analysis was performed in the RMC, IMC and oligomenorrhoea subgroups.

Table IV shows the baseline characteristics categorized by menstrual cycle pattern at the start of the study and the presence of oligomenorrhoea at the end of the study. Girls with oligomenorrhoea at the start of the study who remained oligomenorrhoeic had significantly higher BMI, levels of LH, androstenedione and testosterone, but lower prolactin levels than those who changed to a RMC or IMC pattern. Contrary to our expectations, the mean GI ratio was significantly higher (indicating less insulin resistance) in oligomenorrhoeic girls who remained oligomenorrhoeic compared with oligomenorrhoeic girls who changed to a pattern with a normal cycle length.


View this table:
[in this window]
[in a new window]
 
Table IV. Baseline characteristics of adolescents categorized by menstrual cycle pattern at the start of the study (A) and after 3 years (B)
 
Follow-up of adolescents with RMC at the start of the study
One of the two girls with RMC who developed oligo-amenorrhoea temporarily suffered from secondary hypogonadotrophic amenorrhoea due to anorexia nervosa and was oligomenorrhoeic at the end of the study. The other girl who developed oligomenorrhoea had high androstenedione (10 nmol/l) and dehydroepiandrosterone sulphate (DHEA-S) (13 µmol/l) levels at the start of the study. She had polycystic ovaries at ultrasound. During the follow-up her BMI increased from 19.1 to 26 kg/m2 with an androstenedione concentration of 13 nmol/l, a DHEA-S concentration of 12.0 µmol/l and normal LH and testosterone concentration. She was the only oligomenorrhoeic girl with a GI ratio <4.5 ng/10–4 IU, indicating insulin resistance (4.2 ng/10–4 IU) (Legro et al., 1998Go).

Follow-up of adolescents with IMC at the start of the study
In the IMC subgroup analysis, OC use was a significant factor. The risk of becoming oligomenorrhoeic was lower for those who had started to use OC [OR (95% CI) = 0.2 (0.03–1.0)]. This indicates that more girls might have become oligomenorrhoeic had they not started to use OC. In logistic regression analysis with and without OC users, the risk of becoming oligomenorrhoeic for members of the IMC group decreased with an increase in gynaecological age at the start of the study. This decrease was significant in the complete IMC group and among those with an average length of the menstrual cycle between 35 and 41 days, but not in those with an average length of the menstrual cycle of 21–34 days. Figure 2 shows Kaplan–Meier curves to illustrate the risk of becoming oligomenorrhoeic in adolescents with IMC grouped by average length of the menstrual cycle.



View larger version (14K):
[in this window]
[in a new window]
 
Figure 2. Risk of developing oligomenorrhoea in young women with irregular menstrual cycles with an average length of the menstrual cycle between 22 and 34 days (n = 110) and with an average length of the menstrual cycle between 35 and 41 days (n = 38).

 
In the IMC group, no significant predictive value was documented for body measurements, gonadotrophins, androgens, basal insulin concentrations or GI ratio. Polycystic ovaries at ultrasound were associated with a 5.5-fold (95% CI 1.1–28) higher risk for oligomenorrhoea at age 18 years after adjustment for OC use and gynaecological age. After entering the average length of the menstrual cycle (22–34 days versus 35–41 days) in the model, ultrasound pattern of the ovaries was no longer significant due to the conjunction of an average length between 35 and 41 days and polycystic ovaries at ultrasound.

Follow-up of adolescents with oligomenorrhoea at the start of the study
Endocrine data were available for 47 out of 67 oligomenorrhoeic girls with complete follow-up. Of these 47 girls, 23 (49%) remained oligomenorrhoeic, 12 (26%) had IMC with an average length of the menstrual cycle between 35 and 41 days, 2 (4%) had IMC with an average length of the menstrual cycle between 22 and 34 days and 10 (21%) had RMC.

Table V shows the relationship between clinical, endocrine and ultrasound characteristics of PCOS and the risk for persistent oligomenorrhoea. Modelled as continuous variables in logistic regression, BMI, LH, androstenedione, testosterone and prolactin had a significant predictive value. Dichotomized by the 95th centile of the RMC group obesity, elevated levels of androstenedione or testosterone were associated with a high risk for persistent oligomenorrhoea; however, none of these variables showed a significant predictive value. The power of this analysis was low, due to the low prevalence of oligomenorrhoeic girls with values above this threshold value.


View this table:
[in this window]
[in a new window]
 
Table V. Prevalence of persistent oligomenorrhoea in adolescents with and without various clinical, endocrine and ultrasound characteristics of polycystic ovary syndrome at age 15 years
 
Acne, Tanner stage of pubarche or telarche, waist circumference, hip circumference or waist:hip ratio, basal insulin concentration, GI ratio, DHEA-S, FSH or estradiol had no significant predictive value for persistence of oligo-amenorrhoea.

To evaluate whether oligomenorrhoeic girls with relatively high values of BMI, LH, androstenedione or testosterone were at a higher risk for persistence of oligomenorrhoea, we divided these continuous variables into two groups with a threshold value at the 50th centile of the oligomenorrhoea subgroup. Table VI shows the univariate hazard ratio of remaining oligomenorrhoeic estimated by proportional hazard analysis for those with BMI, LH, androgen or insulin values above the median of the oligomenorrhoea group compared with those with values below the median. In multivariate analysis, none of the hormonal determinants gave a significant improvement of the model above BMI. Figure 3 shows Kaplan–Meier curves comparing the risk of remaining oligomenorrhoeic by gynaecological age in adolescents with a BMI above and below the median of this subgroup.


View this table:
[in this window]
[in a new window]
 
Table VI. Univariate hazard ratio (HR) for persistent oligomenorrhoea after 3 years follow-up of body mass index (BMI), hormone concentrations (n = 47) and ultrasound pattern of the ovaries (n = 40) in girls with oligomenorrhoea at age 15 years
 


View larger version (15K):
[in this window]
[in a new window]
 
Figure 3. Risk for persistent oligomenorrhoea in young women with a body mass index above and below the median (19.6 kg/m2) of this subgroup. Oral contraceptive (OC) users were censored at the gynaecological age at which they started OC.

 
Follow-up of adolescents before or <6 months after menarche at the start of the study
We had no baseline endocrine data of adolescents who were before menarche or <6 months after menarche at the start of the study. Adolescents in these subgroups were oligomenorrhoeic after 1.5 years follow-up in 16 and 13% of cases respectively, and after 3 years follow-up in 13 and 9% of cases respectively. This frequency is significantly higher than the frequency of 6% among girls who were <3 years after menarche at the first phase of the POMP study (P < 0.01). BMI and waist:hip ratio had no significant predictive value in this subgroup.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Predictive value of menstrual cycle pattern
About half of 14–16 year old adolescents with oligomenorrhoea will remain oligomenorrhoeic at least until an age of 18 years, the time at which the mean gynaecological age is 5 years. These findings are in accordance with the study of Southam and Richart (1966Go), who found that oligomenorrhoea with the onset at menarche changed to a normal pattern in ~60% in the following 8 years. This change especially occurred in the first 2 years after menarche. Only 35% of girls who were oligomenorrhoeic 2 years after menarche changed to a normal pattern. In contrast, we found no relation between gynaecological age and the change to a RMC pattern.

Data on the risk of becoming oligomenorrhoeic after an episode of regular or mildly irregular menstrual cycles were lacking. We found that this risk was ~2% for 14–16 year old girls with RMC and 4% for girls with IMC with an average length of the menstrual cycle between 21 and 34 days. Adolescents with IMC and an average length of the menstrual cycle between 35 and 41 days were at increased risk to develop oligomenorrhoea compared with girls with RMC [odds ratio (95% CI): 33 (7–154)]. Since oligomenorrhoea is sometimes defined as an average length of the menstrual cycle of ≥35 days instead of ≥42 days, this subgroup of girls with IMC might also be specified as cases of mild oligomenorrhoea.

Strengths and limitations of our study
The first strength of our study is that it was performed in a sample of the general population without indication for selection bias in the baseline population, providing a good external validity. Furthermore, the ethnicity of the population was homogeneous. Only 10% of the participants were not West Europeans. They were excluded from the analysis. Finally, no previous study had evaluated the predictive value of clinical, endocrine and ultrasound data in a structured way.

Yet the strength of our conclusions could be limited due to 60% of the participants starting with OC. At the time we started the POMP study, reports gave an estimate for OC use at age 18 years of 25–40% (Hirasing, 1987Go). Although we found no evidence for selection bias by menstrual cycle pattern in starting OC, an influence of OC use on the results cannot be completely excluded.

Studies on post-pill amenorrhoea support the idea that the effect on the results of recoding OC users to the menstrual cycle pattern in the year before OC use will be limited. These studies show a persistence of underlying causes of ovulatory dysfunction during OC use and a return of ovulatory dysfunction after discontinuation of OC (Jacobs et al., 1977Go; Taylor et al., 1977Go; Hull et al., 1981Go; Vytiska-Binstorfer et al., 1987Go).

The presented data are particularly valid for adolescents with menarche before age 14.5 years. At the start of the POMP study, we collected endocrine data from girls who were after menarche at that time. In the follow-up study, we found a higher frequency of oligomenorrhoea in the first years after menarche in girls with age at menarche of 14 or 15 years (who were before menarche at the start of the study), 11%, in contrast to those with age at menarche of 12 or 13 years, 6%. We do not know whether the endocrine background and natural course of oligomenorrhoea with early and late menarche was the same.

Relationship between menstrual cycle pattern after menarche and adult oligo-amenorrhoea
After exclusion of the girl who developed anorexia nervosa, the risk of developing normogonadotrophic oligo-amenorrhoea after an episode with RMC is only 1%. However, ~80% of the population have RMC and therefore this subgroup may still substantially contribute to oligomenorrhoea at age 18 years in the general population. Projection of our findings to the baseline population results in an estimate that 13% of all adult patients with oligomenorrhoea have RMC shortly after menarche, 7% IMC with an average length between 21 and 34 days, 23% IMC with an average length between 35 and 41 days and 59% oligo-amenorrhoea from the onset of menarche. Polymenorrhoea was rare in our population and none of these girls developed oligomenorrhoea. The prevalence of hypogonadotrophic secondary amenorrhoea was remarkably low in our study. We had expected a higher frequency of cases with hypogonadotrophic amenorrhoea due to anorexia or bulimia nervosa. However, most of these cases may develop after an episode of RMC and only a sample of those cases were included in our study due to the design.

Adolescent oligo-amenorrhoea and PCOS
As shown in this and other studies, adolescents with oligomenorrhoea have clinical, endocrine and ultrasound signs of PCOS. Yen (1980Go) concluded from case histories of 100 PCOS patients that they presented with a normal mean age at menarche, continuation of post-menarcheal menstrual irregularity, clinically discernible excessive hair growth either before or around the time of menarche and ‘overweight’ compared to patient’s peers prior to menarche (Yen, 1980Go). No prevalence of the various symptoms has been given.

In our population, obesity and hirsutism were rare. Both were associated with an increased risk for persistent oligomenorrhoea. Due to a low prevalence, hirsutism had no significant predictive value in the analysis. Acne is widespread among adolescents and therefore not a specific marker for hyperandrogenism in this age group as it is in adults. Other authors have documented that obesity was less prevalent in adolescents with signs of PCOS compared with adults with PCOS in the same population (Gülekli et al., 1993Go; Dramusic et al., 1997Go). In the Nurses’ Health study II, it was found that a BMI >24.0 kg/m2 at age 18 years was a significant predictor of irregular menstrual cycles or oligomenorrhoea from age 18 to 22 years and anovulatory infertility later in life (Rich-Edwards et al., 1994Go).

Our finding that concentrations of androgens or LH and polycystic ovaries have a predictive value for ovulatory dysfunction later in life agree with earlier studies (Kimura et al., 1988Go; Apter and Vihko, 1990Go; Venturoli et al., 1995Go; Porcu et al., 1997Go). Apter and Vihko (1990Go) showed that higher testosterone concentrations (>1.1 nmol/l) in the first post-menarcheal years were associated with lower fertility in the third decade. Similar to our results, not only androgen concentrations fitting hyperandrogenism were associated with ovulatory dysfunction in adulthood, but also androgen levels in the mid–high normal range. Apter and Vihko gave no attention to the menstrual cycle patterns.

Prevention of PCOS or its symptoms
Our study provides no clues for primary prevention of PCOS. For obese girls, weight reduction is the treatment of first choice for secondary preventions of long-term health consequences of PCOS such as endometrial carcinoma, diabetes mellitus type II, hypertension and cardiovascular disease (Apter, 1998Go; Franks, 2002Go). These long-term consequences seem especially to be related to obesity, insulin resistance and hyperandrogenism (Franks, 1995Go; (Solomon, 1999Go). As we found that a normal BMI was also associated with persistent oligomenorrhoea, weight control may also be effective in this group.

Endometrial hyperplasia should be prevented by treatment with OC or cyclical progestagens. OC with the anti-androgen cyproterone acetate are widely used to prevent the progression of hirsutism. Recently, some concern has arisen about negative effects of cyproterone acetate on dislipidaemia and insulin sensitivity (Creatsas et al., 2000Go; Elter et al., 2002Go). Creatsas et al. (2000) found a similar effect on hirsutism of cyproterone acetate-containing OC compared with those containing desogestrel. Desogestrel containing OC had the advantage of no side-effects on lipid metabolism. Elter et al. (2002Go) showed that adding metformin to cyproterone acetate-containing OC improved insulin sensitivity; no difference was found in lipid profiles.

In the literature, we did not find any evidence that ovulatory dysfunction is influenced either positively or negatively by OC use. However, changes in weight during OC may affect ovulatory function after OC use. There may be concern about OC use-related weight gain; however, it has been documented that weight gain was not different in adolescents who started OC for contraception in comparison with peers who were also sexually active but did not use OC (Carpenter and Neinstein, 1986Go).

In contrast to our findings, Ibanez et al. (2001Go) recently found hyperinsulinaemia in lean, hyperandrogenaemic, oligomenorrhoeic adolescents. They documented a transition to regular menstrual cycles in 78% of these girls after 4 months treatment with metformin. Three months after discontinuation of the metformin treatment, oligomenorrhoea had returned in all girls. They showed that metformin might be helpful to reduce hirsutism and acne, and improve dislipidaemia. Theoretically, the latter may result in secondary prevention of long-term health consequences of PCOS. Although metabolic abnormalities recently have been described in mainly obese adolescents with PCOS (Arslanian et al., 2001Go; Lewy et al., 2001Go; Palmert et al., 2002Go), long-term effects of prescribing insulin sensitizers to insulin-resistant adolescents without glucose intolerance are not known.

In our opinion, prescribing metformin in adolescents just to induce ovulation is not appropriate. When metformin is prescribed, attention should be given to contraception, if fertility is not desired. Although OC may have a slightly negative effect on insulin resistance, it offers adolescents prevention of endometrial hyperplasia and effective contraception.

We conclude that the menstrual cycle pattern in the first years after menarche is a better predictor for ovulatory dysfunction in adulthood than androgen or LH concentrations. Persistent oligomenorrhoea is best predicted by a BMI above the median of their peers who are also oligomenorrhoeic. Becoming oligomenorrhoeic is predicted by IMC, particularly with an average cycle length of 35–41 days and polycystic ovaries. Young women who exhibit these symptoms may be informed about their increased risk for ovulatory dysfunction in their third and fourth decade. Although fertility in individual cases may be uneventful, as a group their fertility will be reduced. They may be reassured that the vast majority of PCOS patients do become pregnant. However, the time to pregnancy is increased (Taylor, 1998Go). If these women desire to have children, this information may guide their decision to discontinue contraception and try to conceive at an earlier age. Attention may be given to the relationship between ovulatory dysfunction and weight.


    Acknowledgements
 
We are indebted to the girls and the management of the schools that participated in the study. We thank the staff of the endocrine laboratory of Vrije Universiteit Medical Center (head Dr C.Popp-Snijders) for performing the hormone determinations.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Anttila L, Koskinen P, Kaihola HL, Erkkola R, Irjala K and Ruutiainen K (1992) Serum androgen and gonadotropin levels decline after progestogen-induced withdrawal bleeding in oligomenorrheic women with or without polycystic ovaries. Fertil Steril 58,697–702.[ISI][Medline]

Apter D (1998) How possible is the prevention of polycystic ovary syndrome development in adolescent patients with early onset of hyperandrogenism. J Endocrinol Invest 21,613–617.[ISI][Medline]

Apter D and Vihko R (1990) Endocrine determinants of fertility: serum androgen concentrations during follow-up of adolescents into the third decade of life. J Clin Endocrinol Metab 71,970–974.[Abstract]

Apter D, Bützow T, Laughlin A and Yen SS (1995) Metabolic features of polycystic ovary syndrome are found in adolescent girls with hyperandrogenism. J Clin Endocrinol Metab 80,2966–2973.[Abstract]

Arslanian SA, Lewy VD and Danadian K (2001) Glucose intoleance in obese adolescents with polycystic ovary syndrome: roles of insulin resistance and {beta}-cell dysfunction and risk of cardiovascular disease. J Clin Endocrinol Metab 86,66–71.[Abstract/Free Full Text]

Carpenter S and Neinstein LS (1986) Weight gain in adolescent and young adult oral contraceptive users. J Adolesc Health Care 7,342–344.[Medline]

Creatsas G, Koliopoulos and Mastorakos G (2000) Combined oral contraceptive treatment of adolescent girls with polycystic ovary syndrome. Lipid profile. Ann NY Acad Sci 900,245–255.[Abstract/Free Full Text]

Dramusic V, Goh VHH, Rajan U, Wong YC and Ratnam SS (1997) Clinical, endocrinologic, and ultrasound features of polycystic ovary syndrome in Singaporean adolescents. J Pediatr Adolesc 10,125–132.

Elter K, Imir G and Durmusoglu F (2002) Clinical, endocrine and metabolic effects of metformin dded to ethinyl estradiol–cyproterine acetate in non-obese women with polycystic ovarian syndrome: a randomized controlled study. Hum Reprod 17,1729–1737.[Abstract/Free Full Text]

Ferriman D and Gallwey JD (1961) Clinical assesment of body hair growth in women. J Clin Endocrinol Metab 21,1440–1447.[ISI]

Franks S (1995) Polycystic ovary syndrome. N Engl J Med, 333, 853–861.[Free Full Text]

Franks S (2001) Adult PCOS begins in childhood. Best Pract Res Clin Endocrinol Metab 16, 263–272.[ISI]

Gardner J (1983) Adolescent menstrual characteristics as predictors of gynaecological health. Ann Hum Biol 10,31–40.[ISI][Medline]

Gülekli B, Turhan NÖ, Senöz S, Kukner S, Oral H and Gokmen O (1993) Endocrinological, ultrasonographic and clinical findings in adolescent and adult polycystic ovary patients: a comparative study. Gynecol Endocrinol 7,273–277.[ISI][Medline]

Hirasing R (1987) Het menstruatiepatroon bij adolescenten in Westfriesland. Thesis University of Groningen, The Netherlands. Van Gorcum, Assen, The Netherlands.

Hull MG, Bromham DR, Savage PE, Barlow TM, Hughes AO and Jacobs HS (1981) Post-pill amenorrhea: a causal study. Fertil Steril 36,472–476.[ISI][Medline]

Ibanez L, Valls C, Ferrer A, Marcos MV, Rodriguez-Hierro F and de Zegher F (2001) Sensitization to insulin induces ovulation in nonobese adolescents with anovulatory hyperandrogenism. J Clin Endocrinol Metab 86,3595–3598.[Abstract/Free Full Text]

Jacobs HS, Knuth UA, Hull MG and Franks S (1977) Post-pill amenorrhea—cause or coincidence? Br Med J 2,940.[Medline]

Kimura K, Minakami H and Tamada T (1988) [A longitudinal study on the prognosis of ovulatory disturbance in teenage patients with high LH and normal FSH serum levels]. Nippon Naibunpi Gakkai Zasshi 64,1088–1101.[Medline]

Legro RS, Finegood D and Dunaif A (1998) A fasting glucose to insulin ratio is a useful measure of insulin sensitivity in women with polycystic ovary syndrome. J Clin Endocrinol Metab, 83, 2694–2698.[Abstract/Free Full Text]

Lewy VD, Danadian K, Witchel F and Arslanian S (2001) Early metabolic abnormalities in adolescent girls with polycystic ovarian syndrome. J Pediatr 138,38–44.[CrossRef][ISI][Medline]

Martins JM, Carreiras F, Afonso A, Falcao J and Charneco daCosta J (1998) Transient hyperandrogenemia and its relation to ovulation. Fertil Steril 70,664–670.[CrossRef][ISI][Medline]

Minakami H, Abe N, Izumi A and Tamada T (1988) Serum luteinizing hormone profile during the menstrual cycle in polycystic ovarian syndrome. Fertil Steril 50,990–992.[ISI][Medline]

Palmert MR, Gordon CM, Kartashov AI, Legro RS, Emans SJ and Dunaif A (2002) Screening for abnormal glucose tolerance in adolescents with polycystic ovary syndrome. J Clin Endocrinol Metab 87,1017–1023.[Abstract/Free Full Text]

Plewig G and Kligman AM (1975) Acne Morphogenesis and Treatment. Springer-Verlag, Berlin.

Porcu E, Venturoli S, Longhi M, Fabbri R, Paradisi R and Flamigni C (1997) Chronobiologic evolution of luteinizing hormone secretion in adolescence: developmental patterns and speculations on the onset of the polycystic ovary syndrome. Fertil Steril 67,842–848.[CrossRef][ISI][Medline]

Rich-Edwards JW, Goldman MB, Willett WC, Hunter DJ, Stampfer MJ, Colditz GA and Mansin JE (1994) Adolescent body mass index and fertility caused by ovulatory disorder. Am J Obstet Gynecol 71,171–177.

Roede MJ and Van Wieringen JC (1985) Growth diagrams 1980: Netherlands third nationwide survey. Tijds Soc Geneeskd 63,1–34.

Rosenfield RL, Ghai K, Ehrmann DA and Barnes RB (2000) Diagnosis of the polycystic ovary syndrome in adolescence: comparison of adolescent and adult hyperandrogenism. J Pediatr Endocrinol Metab 13(Suppl 5),1285–1289.[ISI][Medline]

Siegberg R, Nilsson CG, Stenman UH and Widholm O (1986) Endocrinological features in oligomenorrheic adolescent girls. Fertil Steril 46,852–857.[ISI][Medline]

Solomon CG (1999) The epidemiology of polycystic ovary syndrome. Prevalence and associated disease risks. Endocrinol Metab Clin North Am 28,247–263.[ISI][Medline]

Southam AL and Richart RM (1966) The prognosis for adolescents with menstrual abnormalities. Am J Obstet Gynecol 94,637–645.[ISI][Medline]

Tanner JM (1962) Growth at Adolescence. Blackwell, Oxford.

Taylor AE (1998) Polycystic ovary syndrome. Endocrinol Metab Clin North Am 27,877–901.[ISI][Medline]

Taylor AE, McCourt B, Martin KA, Anderson EJ, Adams JM, Schoenfeld D and Hall JE (1997) Determinants of abnormal gonadotropin secretion in clinically defined women with polycystic ovary syndrome. J Clin Endocrinol Metab 82,2248–2256.[Abstract/Free Full Text]

Taylor RN, Jr, Berger, GS and Treloar AE (1977) Changes in menstrual cycle length and regularity after use of oral contraceptives. Int J Gynaecol Obstet 15,55–59.[Medline]

Treloar AE, Boynton RE, Behn BG and Brown BE (1967) Variation of the human menstrual cycle through reproductive life. Int J Fertil 12,77–126.[ISI][Medline]

van Hooff MH, Hirasing RA, Kaptein MB, Voorhorst FJ, Koppenaal C and Schoemaker J (1998a) The use of oral contraception by adolescents for contraception, menstrual cycle problems or acne. Acta Obstet Gynecol Scand 77,898–904.[CrossRef][ISI][Medline]

van Hooff MH, Voorhorst F, Kaptein M, Hirasing RA, Koppenaal C and Schoemaker J (1998b) Relationship of the menstrual cycle pattern in 14–17 year old adolescents with gynaecological age, body mass index, historical parameters. Hum Reprod 13,2252–2260.[Abstract]

van Hooff MH, van der Meer M, Lambalk CB and Schoemaker J (1999a) Variation of LH and androgens in oligomenorrhoea and its implications for the study of the polycystic ovary syndrome. Hum Reprod 14,1684–1689.[Abstract/Free Full Text]

van Hooff MH, Voorhorst FJ, Kaptein MB, Hirasing RA, Koppenaal C and Schoemaker J (1999b) Endocrine features of the polycystic ovary syndrome in a random population sample 14–16 year old adolescents. Hum Reprod 14,2223–2229.[Abstract/Free Full Text]

van Hooff MH, Voorhorst FJ, Kaptein MB, Hirasing RA, Koppenaal C and Schoemaker J (2000a) Polycystic ovaries in adolescents and the relationship with menstrual cycle patterns, luteinizing hormone, androgens, and insulin. Fertil Steril 74,49–58.[CrossRef][ISI][Medline]

van Hooff MH, Voorhorst FJ, Kaptein MB, Hirasing RA, Koppenaal C and Schoemaker J (2000b) Insulin, androgen, and gonadotropin concentrations, body mass index, and waist to hip ratio in the first years after menarche in girls with regular menstrual cycles, irregular menstrual cycles, or oligomenorrhea. J Clin Endocrinol Metab 85,1394–1400.[Abstract/Free Full Text]

Veldhuis JD, Pincus SM, Garcia-Rudaz MC, Ropelato MG, Escobar ME and Barontini M (2001) Disruption of the joint synchrony of luteinizing hormone, testosterone, and androstenedione secretion in adolescents with polycystic ovarian syndrome. J Clin Endocrinol Metab 86,72–79.[Abstract/Free Full Text]

Venturoli S, Porcu E, Fabbri R, Pluchinotta V, Ruggeri S, Macrelli S, Paradisi R and Flamigni C (1995) Longitudinal change of sonographic ovarian aspects and endocrine parameters in irregular cycles of adolescence. Pediatr Res 38,974–980.[Abstract]

Vihko R and Apter D (1990) Endocrine determinants of fertility: serum androgen concentrations during follow-up of adolescents into the third decade of life. J Clin Endorinol Metab 71,970–974.[Abstract]

Vollman RF (1977) In Friedman EA (ed), The menstrual cycle. (Major problems in obstetrics and gynecology). WB Saunders, Philadelphia. pp. 74–159.

Vytiska-Binstorfer E, Huber JC, Spona J and Gitsch E (1987) [Endocrine profile of patients with post-pill amenorrhea]. Geburtshilfe Frauenheilkd 47,414–416.[ISI][Medline]

Westrate JA, Deurenberg J and Van Tinteren H (1989) Indices of body fat distribution and adiposity in Dutch children from birth to 18 years of age. Int J Obesity 13,465–477.[ISI][Medline]

Yen SS (1980) The polycystic ovary syndrome. Clin Endocrinol 12,177–207.[ISI][Medline]

Submitted on August 15, 2002; resubmitted on August 4, 2003; accepted on October 17, 2003.