Abstract

Objective

Obesity is associated with an increased risk of polycystic ovary syndrome (PCOS). It can be difficult to differentiate between PCOS and physiological oligomenorrhoea/anovulation in adolescent girls. To date, studies of the prevalence of PCOS in adolescents have predominantly used diagnostic criteria validated primarily in adult women. The aim of this study was to investigate the prevalence of PCOS in obese girls using the current diagnostic criteria for adolescents.

Methods

The diagnosis of PCOS was based on the presence of menstrual irregularity, clinical hyperandrogenism and hyperandrogenemia and the exclusion of other causes. Patients with one or two of these conditions were classified as “at risk for PCOS”. The control group consisted of patients with obesity alone but no other comorbidity.

Results

A total of 421 patients were included in the study. The number of patients meeting the definition of PCOS was 35, representing a prevalence of 8.3%, while 200 patients (46%) were defined as “at risk for PCOS”. The diagnostic value of the free androgen index (FAI) was found to be adequate, while other tests were poor. The cut-off values were 11 for FAI and 0.44 ng/mL for total testosterone, with optimal sensitivity and specificity.

Conclusion

Despite the increasing number of studies, the diagnosis and management of PCOS in adolescents remains challenging. While efforts should be made to avoid overdiagnosis, it is also important to recognize that many more patients may be at risk of developing PCOS.

Introduction

Polycystic ovary syndrome (PCOS) is characterized by androgen excess, ovulatory dysfunction and dysfunction of the hypothalamic-pituitary-ovarian axis, resulting in a self-perpetuating vicious cycle of neuroendocrine and metabolic dysfunction. The disease is complex and multigenic, with environmental factors playing a role (1). Obesity and insulin resistance are associated with an increased risk of PCOS and, when present in PCOS, more severe clinical manifestations (2). During puberty, insulin resistance increases due to activation of the growth hormone axis. This leads to compensatory hyperinsulinemia, which increases ovarian and adrenal androgen production, decreases hepatic sex hormone binding globulin (SHBG) production and impairs ovarian follicular dynamics and function (3). PCOS has become an important comorbidity in obese adolescent girls because of the presence of these physiological changes, environmental factors affecting metabolic health and epigenetic factors may also be involved.

In 1990, the National Institutes of Health criteria for PCOS were defined as “hyperandrogenism and/or hyperandrogenemia and oligo-anovulation, and exclusion of other diseases” (4). In 2003, the Rotterdam criteria were established, in which the diagnosis of PCOS requires the presence of two or more of the following criteria: oligo-ovulation/anovulation, clinical or laboratory androgen excess, the presence of multiple ovarian cysts, and the exclusion of other diseases (5). According to the criteria established by the Androgen Excess Society in 2006, PCOS should be diagnosed when ovarian dysfunction and/or polycystic ovarian morphology (PCOM) and clinical and/or biochemical findings of hyperandrogenism were present and other causes were excluded (6).

Due to the physiological transition from menarche to full biological maturity in adolescent girls, it can be difficult to differentiate true cases of PCOS from those who are normal. To date, studies investigating the prevalence and characteristics of PCOS in obese adolescents have predominantly used the diagnostic criteria described earlier, which have been validated primarily in adult women. In 2020, Peña et al. (7) published an international evidence-based guideline for adolescents. This latest guideline gives specific recommendations for diagnosing PCOS in adolescents, thus aiming to avoid misdiagnosis or delayed and under- or over-diagnosis. Further aims were to avoid unnecessary testing and to identify adolescents ‘at risk’ of PCOS. In this guideline, the definition of PCOS in adolescent girls is met by the criteria of “clinical and/or biochemical hyperandrogenism” in the presence of “irregular menstrual cycle” and exclusion of other possible causes. It was also noted that ultrasound findings should not be used for diagnostic purposes and that biochemical hyperandrogenism should be determined using appropriate high-quality assays. An international consortium update by Ibáñez et al. (8) has previously recommended that confirmation of biochemical hyperandrogenism in adolescents is important before a definitive diagnosis of PCOS is made, due to ethnic and racial differences in the clinical signs of androgen excess.

The aim of the present study was to investigate the prevalence of PCOS and the prevalence of being at risk of PCOS in obese adolescent girls in the context of the adolescent-specific diagnostic criteria described by Peña et al. (7). In addition, the study aimed to investigate the clinical and biochemical characteristics of these girls, thereby contributing updated information to the existing literature on this topic.

Methods

The present study was conducted with female patients who presented to the Pediatric Endocrinology outpatient clinic because of excessive weight gain between January 1, 2019, and March 1, 2023. Of the 1,000 patients aged 15-18 years who presented with excessive weight gain, 421 girls with a body mass index (BMI) above the 95^th percentile, who had reached final height and pubertal stage (Tanner-5) and had menarche at least one year prior were included in the study. The study population consisted of individuals who completed the study forms in full and consented to undergo examination of all body parts for hirsutism. Informed consent forms were obtained from the patients themselves. Individuals with a medical history of congenital or acquired systemic diseases, or drug use that could lead to menstrual irregularities and abnormal body hair, were excluded from the study.

A questionnaire was administered to patients inquiring about their age at the onset of menstruation, the frequency of their menstrual cycles, the number of days and amount of bleeding, the presence of acne, and the presence of abnormal hair growth outside the genital and axillary regions.

Obesity was defined as BMI >95^th percentile according to age and gender or >30 kg/m² (whichever was lower) in adolescents who had reached their final height (9). The World Health Organization defines obesity as “mild” if the BMI is between 30 and 35 or greater than 100 and 120% of the 95^th percentile, “moderate” if the BMI is between 35 and 40 or greater than 120 and less than 140% of the 95^th percentile, and “severe” if the BMI is greater than 40 or greater than 140% of the 95^th percentile. In this study, the BMI reference values for Turkish children were employed (10, 11).

The diagnosis of PCOS was made based on the presence of menstrual irregularity, clinical hyperandrogenism (confirmed through biochemical hyperandrogenemia), and the exclusion of other potential causes. In instances where patients exhibited one or two of these conditions but did not fully meet the diagnostic criteria for PCOS, they were classified as “at risk for PCOS.”

Menstrual Irregularity

The patients who had experienced menstrual cycles for at least one year were included in the study. The definition of irregularity used the Peña et al. (7) guideline recommendations:

1. 1-3 years after menarche: inter-cycle interval <21 or >45 days,

2. >3 years after menarche: inter-cycle interval <21 or >35 days or <8 cycles per year,

3. >90 days for any cycle >1 year after menarche,

4. No menstruation at age 15 or 3 years after thelarche (primary amenorrhea).

Clinical Hyperandrogenism

On physical examination, male-type terminal hair in hormone-sensitive areas (Ferriman-Gallwey score >5-6) was considered “hirsutism” (12), and moderate and/or severe inflammatory-nodular acne on the face, back, and extensor faces with more than 10 lesions on the face was considered “severe acne” (13). Hirsutism and/or severe acne were accepted as clinical hyperandrogenism.

Hyperandrogenemia

The total testosterone and free androgen index (FAI) [total testosterone (nmol/L) / SHBG (nmol/L) x100] parameters were used. Blood samples were collected within the first week of the follicular phase. In our laboratory, total testosterone is quantified by means of an electrochemiluminescence immunoassay method, employing the Roche E801 autoanalyzer and Roche branded test kits. SHBG concentration was also determined by an electrochemiluminescence immunoassay method, using Elecsys SHBG kits on a Cobas e immunoassay analyzer. A value of >0.48 ng/mL for total testosterone, which is the upper limit of the adult female reference value provided by our test kit, and a value of >5 for FAI according to the recent literature were accepted as biochemical hyperandrogenemia (14). In the presence of clinical hyperandrogenism, total testosterone >0.48 ng/mL and/or FAI >5 was considered hyperandrogenism.

Polycystic Ovarian Morphology

In patients in whom ultrasound was required to support the diagnosis, the appearance characterized by multiple peripheral cysts with any ovarian volume above 12 mL was considered “PCOM”. Ultrasonography findings were not used for diagnostic purposes because they were not appropriate for the gynecologic age of our patient group.

Insulin Resistance

A baseline homeostatic model assessment of insulin resistance (HOMA-IR) value of >3.8 in girls or elevated serum insulin levels (>300 µU/mL) on oral glucose tolerance test, with/without impaired glucose metabolism, was considered “insulin resistance” in the presence of clinical findings, such as truncal obesity, keratosis pilaris and acanthosis nigricans (15).

Exclusion of Other Causes

Among patients with 17-hydroxyprogesterone (17-OHP) levels >2 ng/mL, those who demonstrated normalization on repeated measurements or normal results on the adrenocorticotropin hormone stimulation test were included in the study group. Patients with dehydroepiandrosterone sulfate (DHEAS) levels >600 µg/dL underwent ultrasonography, and those with no adrenal mass were included in the study. Among patients with prolactin levels >25 ng/mL, a clinical evaluation was conducted, and those whose prolactin levels normalized on repeated measurements in the follow-up were included in the study group. Those whose prolactin levels continued to increase were excluded. Patients with abnormal thyroid hormones at the time of diagnosis and for three months prior were excluded. Patients receiving thyroid hormone replacement therapy and euthyroid within the last three months were included in the study group. Patients receiving corticosteroids, antipsychotics, or antidepressant drugs were excluded from the study.

The control group consisted of patients with obesity alone, without menstrual irregularities, clinical or biochemical primary hyperandrogenism, or any other comorbidity. Patients with insulin resistance and metformin users were also excluded from the control group.

Statistical Analysis

Data were analyzed using Statistical Package for the Social Sciences, version 25.0 (IBM Inc., Armonk, NY, USA).

The distribution of the data was evaluated using the Kolmogorov-Smirnov test. Descriptive statistical methods were employed to evaluate the study data. Data that conformed to a normal distribution are expressed as the mean and standard deviation, while those that did not conform are expressed as the median and interquartile range. The independent samples t-test was employed for the comparison of normally distributed data between two groups, while the one-way ANOVA test was used for the comparison of more than two groups. The Mann-Whitney U test was used for comparison of non-normally distributed data between two groups, and the Kruskal-Wallis test was used for comparison of more than two non-parametric groups. If the results were found to be significantly different in more than two group comparisons, the post-hoc Bonferroni test was applied. Pearson correlation analysis was employed for data that exhibited a normal distribution, whereas Spearman correlation analysis was used for data that did not. The objective was to ascertain the relationship between the study parameters. The chi-square test was used to evaluate categorical data. The significance level was set at p<0.05 for all tests.

Receiver operating characteristic (ROC) curve analysis was employed to determine the sensitivity and specificity ratios, the area under the curve (AUC), the value of the diagnostic test, and the optimal cut-off values for diagnostic and supportive biochemical markers. Tests with an AUC value greater than 0.6 were considered “usable but inadequate,” while those with an AUC value greater than 0.7 were considered “satisfactory”.

Results

The mean age of the 421 adolescent female patients included in the study was 16.29±0.94 years. The gynecologic age (post-menarche period) was between 1 and 8 years, with a mean of 4.01±1.41 years. The age at which menstruation first occurred ranged from 8 to 17.5 years, with a mean of 12.27±1.24 years. Twenty-four patients (5.7%) had menstruated at the age of 10 years or younger, while 14 patients (3.3%) had started menstruating at the age of 15 years or older.

In our cohort of obese adolescent girls, 126 (29.9%) patients exhibited menstrual irregularities, 139 (33.0%) demonstrated clinical hyperandrogenism, and 120 of 157 (76.4%) patients with laboratory tests exhibited biochemical hyperandrogenism. Of the 139 patients with clinical hyperandrogenism, 83 (59.7%) exhibited hirsutism alone, 21 (15.1%) exhibited severe acne, and 35 (25.2%) exhibited both hirsutism and severe acne. Of the 112 patients who underwent pelvic ultrasound examination, 57 (50.9%) were found to have PCOM. Acanthosis nigricans was identified in 144 (34.2%), and 81 (56.25%) of them were diagnosed with insulin resistance supported by laboratory findings, and metformin treatment was initiated in addition to lifestyle change recommendations.

The number of patients meeting the definition of PCOS was 35, representing an 8.3% prevalence. Moreover, 200 patients (46%) were defined as “at risk for PCOS” based on the presence of one or two of the following criteria: menstrual irregularities, clinical hyperandrogenism, or biochemical hyperandrogenism. Among the 35 patients diagnosed with PCOS, 14 (40%) had been prescribed metformin due to insulin resistance, while five (14.3%) had a family history of type 2 diabetes mellitus. Among the 200 patients considered to be at risk of PCOS, 45 (22.5%) had been prescribed metformin due to insulin resistance, while 47 (23.5%) had a family history of type 2 diabetes mellitus.

There were no statistically significant differences in age, age at menarche, gynecological age, or anthropometric measurements between the PCOS, PCOS at risk, and control groups. Fasting insulin, HOMA-IR, and low-density lipoprotein (LDL) were higher in those at risk for PCOS compared to the control group (p=0.009, p=0.010, and p=0.043, respectively). In addition, aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were higher in PCOS group than in control group (p=0.026 and p=0.006, respectively).

The levels and comparison of hormone tests of patients in the “PCOS” and “at risk” groups are shown in Table 1. The results indicated that, except for biochemical hyperandrogenism and insulin resistance, there were no significant differences in clinical presentations according to the severity of obesity (Table 2).

In correlational analyses, a positive correlation was observed between total testosterone and luteinising hormone/follicle stimulating hormone (LH/FSH) ratio, as well as 1,4-androstenedione (r=0.274; p<0.01; r=0.493; p<0.001, respectively). The FAI demonstrated a positive correlation with total testosterone, 1,4-androstenedione, and LH/FSH ratio (r=0.609; p<0.001, r=0.534; p<0.001, and r=0.292; p<0.001, respectively). Furthermore, the LH/FSH ratio demonstrated a positive correlation with the inter-cycle interval (r=0.364; p=0.003).

Data were analyzed from 167 patients (35 with PCOS and 132 at risk for PCOS) in whom the parameters total testosterone, FAI, 1,4-androstenedione and LH/FSH ratio were measured as diagnostic markers. The AUC values in the ROC curve were significant and fell into the “usable” category for all tests, but their diagnostic values were low. Among the diagnostic tests, FAI was classified as adequate (AUC=0.703, p<0.001), while the AUC values of the other tests were poor (Figure 1).

Furthermore, the cut-off value for the FAI was 11, with an optimal sensitivity and specificity of 59% and 72%, respectively. The cut-off value for total testosterone was 0.44 ng/mL, with an optimal sensitivity and specificity of 76% and 60%, respectively.

Although not used for diagnostis but as a supporting and follow-up parameter, the cut-off value for 1,4-androstenedione was 3.5 ng/mL with an optimal sensitivity of 82% and specificity of 56%. The cut-off for the early follicular phase LH/FSH ratio, a parameter known to increase in chronic anovulation but not used for diagnostic purposes, was found to be >2 with an optimal sensitivity of 59% and specificity of 60%.

Discussion

The prevalence of obesity and, in parallel, the frequency of PCOS, one of the most important comorbidities of obesity in adolescent girls, is increasing. Environmental and epigenetic changes may also increase the propensity for PCOS independently of obesity. The present study was conducted to provide updated information on the prevalence of PCOS in accordance with current diagnostic criteria, clinical features, and predisposing factors in Turkish adolescent girls with obesity. The results indicated that the prevalence of PCOS, defined as clinical hyperandrogenism (hirsutism and/or severe acne) that has been biochemically confirmed (early follicular phase total testosterone >0.48 ng/mL and/or FAI >5) in the presence of irregular menstrual cycles, was 8.3% in obese Turkish girls aged 15-18 years. The prevalence of individuals classified as “at risk for PCOS,” defined as having one or two of these conditions but not meeting the criteria for PCOS, was 46%.

The diagnosis of PCOS during adolescence may be challenging due to the complex nature of the physiological transition period, individual variability, and ethnic and environmental factors among populations, and the frequent overlap between normal and pathological conditions. Treatment approaches also vary in this age group. This study considered the current recommendations for the diagnosis of PCOS, including irregular menses and a rational evaluation of hyperandrogenism due to the features of the population.

For many years the Rotterdam criteria were used to diagnose PCOS in adolescents, prior to the publication of a consensus document by Ibáñez et al. (8) in 2017. However, more recent data suggest that the presence of PCOM in an adolescent who does not have hyperandrogenism and oligo-anovulation does not indicate a diagnosis of PCOS. In 2020, Peña et al. (7) published a guideline that built upon the evidence-based international guidelines on adolescent PCOS by Teede et al. (16) in 2018. This guideline offers further recommendations to enhance diagnostic accuracy and prevent overdiagnosis. The Peña et al. (7) guideline defined irregular menstrual cycles according to the gynecologic age. Irregular menstrual cycles ≤1 year post-menarche represent a normal pubertal transition. The definition of hyperandrogenism required the presence of hirsutism and/or severe acne, while the definition of biochemical hyperandrogenism required measurements using validated, high-quality assays. Pelvic ultrasound was not recommended for the diagnosis of PCOS within 8 years post menarche, and anti-Müllerian hormone levels were not recommended for PCOS diagnosis. Peña et al. (17) also published a literature review on the diagnostic criteria in 2022 defining PCOS as irregular menstrual cycles and hyperandrogenism (clinical and/or biochemical) after excluding other conditions that mimic PCOS. In the present study, irregular menstruation was defined as described in the guideline by Peña et al. (7) and clinical hyperandrogenism was defined as severe acne resistant to topical therapy and/or a Ferriman Gallwey score of >6. In adult guidelines and current adolescent criteria, androgen excess is defined using clinical findings and/or biochemical tests. However, since there are no ultra-sensitive methods for androgen measurements and again considering the genetic characteristics of our population, we considered the recommendation that clinical hyperandrogenism findings should be confirmed with laboratory tests (8). Following these recommendations, the diagnosis of PCOS in the present study was made in cases of irregular menstruation and clinical hyperandrogenism, together with a total testosterone level of >0.48 ng/mL (the upper limit for adult woman given by our laboratory) and/or FAI of >5.

The results of our study once again showed that the sensitivity and specificity of diagnostic tests are not high and that none of the tests used in diagnosis are perfect markers.

In a study conducted in Turkiye, clinical hyperandrogenism was detected in 44 (84.6%) while biochemical hyperandrogenism was detected in 40 (76.9%) of 52 adolescents with PCOS (18). Another study by Yüce et al. (19) showed clinical hyperandrogenism in 76 (67.9%) of 112 PCOS adolescent patients, while biochemical hyperandrogenism was detected in 30 (30.9%). It is thought that the variability between clinical hyperandrogenism rates and biochemical hyperandrogenism rates in different studies may be related to differences in measurement methods and accepted cut-off values. In addition, since hirsutism is determined clinically by observational scoring, it is a highly subjective assessment. There may also be differences in the evaluation of acne as androgen excess.

Despite the control group comprising obese adolescents, fasting insulin, HOMA-IR, and LDL parameters, biomarkers of possible insulin resistance or metabolic syndrome, were higher in those deemed at risk for PCOS than in the control group. Moreover, ALT and AST, which are the markers used to screen for fatty liver, were higher in PCOS patients than in controls. Our findings are consistent with the evidence that insulin resistance and metabolic syndrome in obese adolescents are more strongly associated with chronic anovulation and hyperandrogenism, and that hyperinsulinism plays an important role in the pathogenesis of PCOS (20, 21, 22, 23). Obesity-associated hyperandrogenemia also occurs, due to expanded adipose tissue and potential effects of abnormal adipokine/cytokine levels (24). An increased risk of cardiovascular disease has been reported in women with PCOS, regardless of the severity of obesity. Due to differences in body composition and genetics, individuals with similar BMIs have different levels of systemic inflammation, insulin resistance and metabolic dysfunction. Studies have shown that PCOS is more common in obese women who have an unhealthy metabolic profile (25).

Correlation analyses indicated a positive correlation between total testosterone, 1,4-androstenedione, and the LH/FSH ratio, which supports the hypothesis of chronic anovulation. The elevated androstenedione and 17-OHP levels observed in the PCOS group, in the absence of a difference in DHEAS levels, suggest that the increased androgens are of gonadal origin. However, none of the diagnostic tests was found to be highly predictive, with the FAI being the most reliable among them.

Study Limitations

As our study did not include a healthy control group in which diagnostic tests were performed, the cut-off values we determined were those that differentiate between individuals “at risk of PCOS” and individuals diagnosed with “PCOS.” Since hyperandrogenism alone or menstrual irregularity alone is not diagnostic, in our clinical practice, biochemical tests are usually ordered if both are present, or the signs of clinical androgenism is significant. The majority of patients with only irregular menses did not undergo hormone tests, so only 132 of 200 patients who were defined as “at risk” were involved in the analyses for cut-off values. In addition, the subjective evaluation of the clinical findings required for the diagnosis, which constitutes a limitation in almost all PCOS studies in adolescents, was also a limitation of our study.

Conclusion

The diagnostic value of laboratory tests usually depends on the presence of clinical criteria. It is therefore suggested that clinical findings not supported by laboratory tests should not be considered diagnostic of PCOS. Despite the increasing number of scientific studies and the constant updating of guidelines, the diagnosis and management of PCOS in adolescents remains a puzzle. While efforts should be made to avoid overdiagnosis, it is also important to recognize that many more patients may be ‘at risk’ of developing PCOS. It is recommended that obese girls at risk should be closely monitored for progressive findings. The most recent publication on diagnostic criteria by Ibáñez and de Zegher (12) in 2023 proposed a diagnostic approach by combining criteria from a consortium rooted in pediatric endocrinology and criteria from a consortium rooted in adult endocrinology and gynecology. These recent combined criteria and the cut-offs for diagnostic tests identified in the present study may be incorporated into the methodology of future studies in Turkish adolescent girls with obesity.