Key diagnostic criteria for adolescent polycystic ovary syndrome adjusted for age-specific hormonal and metabolic status standard

Polycystic ovary syndrome (PCOS) is estimated to affect 2.2–7.5% of the adolescent population and up to 68% of women with menstrual irregularities and hirsutism [1, 2]. Severe reproductive and cardiovascular complications of the disease necessitate early diagnosis and correction of PCOS-associated conditions during adolescence before the development of its severe clinical manifestations.

Currently, there are no validated diagnostic criteria for PCOS in adolescents, and the available recommendations permit only to suspect the diagnosis [2, 3]. The 2015 American Association of Clinical Endocrinologists (AACE), the American College of Endocrinology (ACE), and Androgen Excess and PCOS Society (AES) guidelines, as well as the 2014 European Society of Endocrinology and the 2013 American Society of Endocrinology guidelines, were developed primarily for the management adult patients and, as the authors themselves indicate, they require revision and adaptation for adolescent patients [3–5].

PCOS has an extremely diverse clinical presentation. In adolescence, the symptoms of PCOS may be similar to those accompanying normal puberty [1, 2]. In the first years after the menarche, adolescents often have irregular anovulatory cycles. In the first post-menarche year, 85% of the cycles are anovulatory, and for 59% of the cycles, this trend continues after three years [4, 6]. Moreover, in studies involving adolescents, it has been shown that the length of the menstrual cycle less than 20 days or persistent oligomenorrhea with the length of the menstrual cycle more than 45 days two years after menarche could be considered as a sign of oligo-anovulation [5, level of evidence B].

Hormonal changes in puberty include the pulsatile release of hypothalamic gonadotropin-releasing hormone, physiological hyperinsulinemia, and functional hyperandrogenism (HA), leading to acne and hirsutism [3, 4]. Mild hirsutism and isolated acne and /or alopecia alone in adolescence cannot be considered clinical evidence for HA [6, level of evidence B].

Hirsutism in adolescents is difficult to evaluate, including grading of hirsutism based on the Ferriman-Gallway scoring system (1961), which is not adapted for children and is not recommended for the use by the Androgen Excess and PCOS Society [6, 7].

Analysis of hormonal profile and testing for biochemical HA in adolescents has its difficulties due to the absence of reference ranges of hormones for girls with normal puberty. In the Russian Federation, reference standardized laboratory normative indicators are used for girls aged 14.5–15.5, 15.5–16.5, and 16.6–17.5 years [8]. However, the reference ranges ​​for healthy contemporary adolescents are currently changing, and current values need updating to be integrated into clinical practices.

One of the most sensitive methods to assess biochemical hyperandrogenism as a diagnostic marker of PCOS is the free androgen index (FAI), which is the ratio of the total testosterone level and the sex hormone-binding globulin (SHBG) level [2, 6]. A low level of SHBG in itself is one of the markers of PCOS, interrelated with insulin resistance and HA, including in adolescence [7]. For the diagnosis of HA, the AES working group validated testing for androstenedione level, which is a testosterone precursor that does not bind to SHBG; testing for androstenedione provides higher specificity and sensitivity than that for testosterone [7, 9]. However, the determination of HA in adolescents has its difficulties, since hormone levels in girls with PCOS are often within the normal adult ranges​​, and tests for blood level of circulating testosterone have significantly limited sensitivity [4].

One of the generally accepted diagnostic criteria for PCOS is polycystic ovarian morphology assessed by ultrasound examination (US) [5]. However, the diagnosis of polycystic ovarian morphology in adolescence is difficult, especially in obese patients [10, 11]. Testing for anti-Müllerian hormone (AMH) has been proposed as a non-invasive screening test for PCOS in addition to ultrasound examination [4, 10]. A 2016 study reported that AMH cutoff value of 7.03 ng/mL had the best diagnostic accuracy with 50.0% and 70.8% specificity and sensitivity for the diagnosis of PCOS in young patients [11].

Therefore, despite numerous studies, there still is no consensus on diagnostic criteria for PCOS in adolescence and threshold values ​​for laboratory and instrumental methods.

The study aimed to clarify the diagnostic criteria and clinical and laboratory features of PCOS in girls aged 15to17.

Materials and methods

The study comprised 130 girls aged 15 to 17 years with PCOS diagnosed by the Rotterdam criteria. The study was approved by the Research Ethics Committee of the V.I. Kulakov NMRC for OG&P. Inclusion criteria for study patients were age of 15 to17 (2-3 years after menarche); the diagnosis of PCOS based on at least two of the following three Rotterdam criteria: oligo-ovulation or anovulation, clinical or biochemical evidence of hyperandrogenism, and polycystic ovaries on ultrasound assessment; exclusion of all other endocrinopathies; no drug therapy in the last 3 months before enrollment into the study, including combined oral contraceptives (COCs); patient informed consent to be included in the study. Exclusion criteria were as follows: hyperprolactinemia; congenital adrenal cortex dysfunction; thyroid disorders; Cushing’s syndrome and disease; pelvic tumors; exacerbation of chronic and acute somatic diseases; genetic syndromes and malformations.

The comparison group consisted of 30 healthy girls of a similar age assigned to health group 1, who had a regular menstrual cycle and no gynecological and endocrine disorders.

All study participants underwent a general clinical examination including a detailed patient medical history, complaints, anthropometric measurements [height, body mass index (BMI), waist-to-hip ratio (WHR)], and the assessment of hirsutism and puberty.

All study participants were tested for the blood biochemistry and lipid profile including total cholesterol, triglycerides (TG), low-density lipoproteins (LDL), high-density lipoproteins (HDL), highly sensitive C-reactive protein, and leptin. Testing was carried out using photometric and turbidimetric methods on automatic analyzers BA-400, A-25 with Biosystems reagents (Spain). A two-hour, 75-gram oral glucose tolerance test (OGTT) was performed 12–16 hours after the last meal. The level of glucose and immunoreactive insulin was determined in fasting whole venous blood and at 120-minute time point of OGTT. Homeostatic model assessment for insulin resistance (HOMA-IR) was calculated. For an indirect assessment of abdominal adipose tissue, the visceral adiposity index (VAI) recommended in wide practice was used. VAI was calculated by the following formula: (waist circumference (cm)/(36.58+ (1.89xIMT (kg/m2))) x (TG (mmol/L)/0.81) x (1.52 / HDL (mmol/L) )).

All patients on days 3-4 of spontaneous menstrual cycle or uterine bleeding after progesterone challenge test in patients with a long delay of menstruation underwent testing for luteinizing hormone (LH), thyroid-stimulating hormone (TSH), follicle-stimulating hormone (FSH), dehydroepiandrogen sulfate (DHEA-S), androstenedione, prolactin (PRL), estradiol (E2), cortisol, testosterone (T), SHBG, and free thyroxine (T4). Hormone concentrations were measured by electro- and immuno-chemiluminescence methods on automatic analyzers Cobas e 411 (F. Hoffmann-La Roche, Switzerland), Immulite 2000, Immulite 1000 (Siemens, USA) using reagents of the same manufacturers. AMH, 17-OH progesterone, SHBG were analyzed by enzyme-linked immunosorbent assay on DYNEX DSX System analyzers and using the DPC system (USA) on an Immulite instrument. The above laboratory equipment has been recommended by the international community to obtain reliable results. The given reference values ​​are validated for Russian adolescents aged 15 to17.

All girls underwent ultrasound examination of pelvic organs, mammary, and thyroid glands on days 3-5 of the menstrual cycle. Ultrasound assessment was carried out using a Vivid q Ultrasound Scanner (GE Healthcare Siemens, Germany) using a linear and convex sensor with a frequency of 1.8-6.0 MHz. The study was performed with a bladder-filled transabdominal method.

The ovarian-uterine index (OUI) was calculated by the formula (VN Demidov, 1990): OUI = 0.5 (0.5 * V right ovary + 0.5 * V left ovary) / uterine thickness [12].

Statistical analysis was performed using the Statistica 8 software (Statsoft Inc.). Categorical variables were reported as counts and proportions (%) and compared using the χ² test. Quantitative variables showing normal distribution were expressed as means (M) and standard deviation (SD) and presented as M (SD); otherwise, the median (Me) and the quartiles Q1 and Q3 in the Me (Q1; Q3) format were reported. Differences in mean values of the variables showing normal distribution were assessed using the Student t-test. If distribution of the data was not normal the nonparametric Mann-Whitney U-test was used.

The adjusted odds ratio (OR) for the development of PCOS under the influence of various risk factors was evaluated using logistic regression with a 95% confidence interval (CI). The diagnostic accuracy of the tests was assessed by multivariate analysis with the construction of ROC curves, calculating the area under the curve and determining the quality of the models.

Results

According to multivariate analysis, significant anamnestic risk factors for the development of PCOS were the girl’s mother’s history of menstruation disorders, PCOS and endocrine infertility (p = 0.0157, aOR= 4.97; 95% CI = 1.33; 18,53); neurological disorders such as intracranial hypertension, impaired muscle tone in the neonatal age and the first year of life (p = 0.0391, aOR= 8.95; 95% CI = 1.09; 73.41); a history of traumatic injury in childhood (p = 0.0233, aOR= 2.82; 95% CI = 1.13; 7.04); repeated delays of menstruation for more than 90 days with menarche (p = 0.0092, aOR= 2.98; 95% CI = 1.28; 6.91). Mothers of adolescent girls with PCOS and metabolic disorders were more likely to have a history of labor and delivery complications than mothers of girls with PCOS without metabolic disorders and healthy girls (p = 0.0220).

Other risk factors - family history of hereditary diseases, a history of somatic, infectious, and allergic diseases in childhood did not have a significant effect.

Patients with PCOS had higher BMI (22.4 (19.9-27.2) versus 20.2 (18.4-21.8), p = 0.0002), a larger waist circumference (75.0 (69.0-85.0) versus 66.0 (62.0-70.0), p = 0.0003) and hip circumference (98.0 (92.0-103.0) versus 0.72 (0.68-0.76), p = 0.0019) than patients in the control group. Multivariate analysis confirmed that BMI (p = 0.0156, OR = 1.08; 95% CI = 1.01; 1.56) and waist circumference (p = 0.0237, OR = 1.11; 95% CI 1.01; 1.19) were significant risk factors for developing PCOS.

As expected, 92 (70.8%) girls with PCOS developed a male pattern of hair growth that results in excess amounts of coarse body hair in areas such as the chin and upper lip, around the nipples, abdomen and inner thighs, while no cases of hirsutism were observed in the control group. More than a half patients with PCOS (77; 59.2%) complained of acne and oily skin, while in the control group only four girls (13.3%) reported these symptoms (p <0.0001, χ2 test).

At the time of the examination, oligomenorrhea was observed in 60 (46.2%) patients, including menstruation delays followed by abnormal uterine bleeding in 17 (13.1%) patients. Amenorrhea was observed in 43 (33.8%) girls, and 5 (3.8%) patients had primary amenorrhea. A regular cycle in the year of the examination was observed in 7 (5.3%) girls with PCOS; however, 4 (3.1%) of them had a history of oligomenorrhea.

Blood lipid profiles did not differ statistically between the groups, but patients with PCOS had a higher VAI (1.0 (0.6-1.6) versus 0.7 (0.5-1.0), p = 0.0461), which indicates an increased risk of cardio-metabolic disorders in patients with PCOS starting in adolescence.

There were significant differences in many parameters of the hormonal profile between patients in PCOS and the control groups; however, they were within laboratory reference ranges and approved age-specific values. Compared with the approved age-specific hormone standards for girls (N. D. Fanchenko et al., 1986 [8]), a change in contemporary indicators was found in “healthy” girls, many of which showed a decrease in LH levels (12; 40.0%) and in the majority - a decrease in the concentration of T (17; 56.7%), which complicates the use of the previous age-specific reference ranges in contemporary adolescents. Our data were validated for a group of healthy Russian adolescents according to the recommendations of international communities on hormonal studies, which allowed us to review the reference values ​​of hormonal parameters in girls aged 15 to17 in the follicular phase on days 2-3 of a spontaneous menstrual cycle (Table 1).

A comparison of hormonal profiles of girls with and without PCOS is shown in table. 2.

The search for threshold values ​​of hormonal parameters for the diagnosis of PCOS in girls aged 15 to17 showed that the most significant criteria for PCOS with high sensitivity and specificity (> 75.0% and 89.0– 93.0%) were AMH> 7.20 ng/ml and FAI > 2.75 (Fig. 1). High sensitivity (63.2–78.2%) and specificity (84.4–93.7%) for the diagnosis of PCOS in girls aged 15 to17 were found for T>1.15 nmol/L, androstenedione>11.45 ng/ml, and LH/FSH ratio>1.23.

Among characteristics of ovarian morphology estimated by 2D pelvic ultrasound the most significant parameters for the diagnosis of PCOS in adolescents were the mean ovarian volume (Vovary mean) > 10.70 cm3, sensitivity 83.0%, specificity 83.0 %, AUC = 84.8%, p <0.05) and the ovarian-uterine index (OUI) (with OUI > 3.95, sensitivity 81.0%, specificity 83.0%, AUC = 83.7%, p <0.05).

Our regression analysis showed that using more than four variables results in the highest diagnostic accuracy for PCOS in adolescents (> 90%); inclusion of more than three variables provides the accuracy of 85%; the use ≤ 2 variables leads to a decrease in the accuracy (Fig. 2). It turned out that of all the variables, AMH, which has a high but insufficient accuracy of 78.7% when used as a single variable. But its combination with leptin (adipokine) results in the maximum diagnostic accuracy (91.8%) and can be recommended for determination (χ2 (2) = 58.3, p <0.00001). The equation for the diagnosis of PCOS is: logit = 85.73–1.73 * [AMH] –0.12 * [Leptin]. When the value of the equation is less than 70.72, the diagnostic sensitivity was 92.5% and specificity 90.0%. Therefore, serum concentrations of AMH and leptin can be used with high accuracy to diagnose PCOS in adolescents.

Discussion

Diagnostic criteria for PCOS in adolescence remain a subject of debate [4, 7]. In 2003, a group of experts from the European Society of Human Reproduction and Embryology (ESHRE) and the American Society for Reproductive Medicine (ASRM) in Rotterdam formulated the most widely accepted diagnostic criteria for the diagnosis of PCOS, which is based on at least two of the following three criteria: oligo-ovulation or anovulation, clinical or biochemical evidence of hyperandrogenism, and polycystic ovaries on ultrasound assessment [2]. However, hormone levels for determining hormonal parameters and biochemical HA in adolescents have not been determined due to their high variability, including variability associated with the use of different laboratory equipment and reagents. We have proposed levels of hormonal indicators validated for the population of Russian adolescents on days 2-3 of the first phase of the menstrual cycle at the age of 15to17, inclusively, determined on the most commonly used equipment for electro- and immuno-chemiluminescence methods on automatic analyzers Cobas e 411 (F. Hoffmann-La Roche, Switzerland), Immulite 2000, Immulite 1000 (Siemens, USA), DYNEX DSX System using reagents from the same manufacturers.

The analysis of the hormonal profile showed that patients with PCOS had significantly higher levels of LH, LH/FSH ratio, T, FAI, 17-OH-progesterone, DHEA-S, androstenedione, and AMH, which is consistent with data reported by many authors [3]. In our study, similarly to the data of Güdücüa N. et al., FAI in patients with PCOS had positive correlations of with BMI (r = 0.46; p <0.05), WHR (r = 0.35; p < 0.05), HOMA-IR (r = 0.30; p <0.05), VAI (r = 0.40; p <0.05), and negative correlation with HDL (r = -0.29; p <0.05), which were not significant in the group of “healthy” girls [13]. These data confirm a positive association of HA of ovarian origin with an atherogenic lipid profile, insulin resistance, and the risk of cardiovascular complications in patients with PCOS already in adolescence. One of the significant markers of PCOS interrelated with insulin resistance and HA is a decrease in the level of SHBG [3, 14]. In our study, in patients with PCOS, the level of SHBG was significantly lower than in the control group (37.0 nmol/L (24.7–55.5) versus 52.9 nmol/L (39.0–67.6); p <0.05), which resulted in a higher level of FAI (5.5 (2.8–7.0) versus 1.6 (1.1–2.3); p <0.0001). In a 2016 study by Yetim A. et al. [2] including 53 patients with PCOS aged 15–20 years and 26 healthy girls, the median level of SHBG in patients with PCOS was significantly lower (25.8 nmol/L versus 49, 6 nmol/L; p <0.01) and the median FAI was significantly higher (6.8 versus 3.0; p <0.0001) than in healthy girls, which is comparable with our data [1].

A 2015 systematic review by Tsikuras P. et al. [2] emphasized the difficulty in diagnosing PCOS in adolescence, which is associated not only with the heterogeneous clinical presentation but also with the absence of definite values of hormonal parameters and HA in adolescents. The authors summarized the data of 24 studies and found that compared with existing reference intervals, in adolescents with PCOS the levels of T, DHEA-S, LH, LH/FSH, and androstenedione were above normal ranges in slightly more than half of the cases, and often were within normal intervals. In our study (based on age-specific normative values for adolescents), the most significant in the diagnosis of hormonal disorders in PCOS were elevated levels of AMH (75.4%), LH (62.3%) and androstenedione (66.2%), i.e., in slightly more than in half of the cases. All other parameters differed from the standards in less than half, and even in a third of the cases. Therefore, the likelihood of confirming the diagnosis is not high enough.

Most authors emphasize the importance of calculating FAI and determining androstenedione in assessing HA in adolescence [2, 7]. Taking into account that in adolescents with PCOS, HA is often within the normative values ​​for adults, and methods for analyzing serum T have significantly limited sensitivity, the authors emphasize the need for studies investigating age-specific threshold levels of hormones in PCOS, as well as determination of reference intervals for healthy girls with normal puberty [3]. A 2015 study by Pinola M.et al. showed that the level of androgens in PCOS decreases with age. At the same time, it was demonstrated that androstenedione levels>9.65 ng/ml, FAI>2.34 [9] had high sensitivity and specificity for the diagnosis of PCOS in women under 40. Our findings of 130 girls with PCOS and 30 “healthy” girls suggest that the use of T>1.15 nmol/L, androstenedione>11.45 ng/ml, LH/FSH ratio>1.23 is appropriate for a comprehensive diagnosis of PCOS in this age range due to a high sensitivity of 63.2–78.2% and a specificity of 84.4–93.7% of these tests. The currently available literature is lacking similar studies in adolescent patients.

The Worldwide Pediatric Consensus (2015) emphasized the feasibility of using AMH as a non-invasive screening test in the diagnosis of PCOS [11]. In a 2016 study of 15–20-year-old girls with PCOS, Yetim A. et al. calculated the cutoff threshold for the level of AMH and reported that AMH level> 6.10 ng/ml had 92.3% sensitivity and 81.1% specificity for the diagnosis of PCOS [1]. In another 2017 study, Merino P.M. et al. after examining 102 adolescents with PCOS, recommended the use of a threshold of AMH> 7.03 ng/ml with 50.0% sensitivity and 70.8% specificity [11]. However, the number of studies investigating the threshold values of hormones for the diagnosis of PCOS in adolescence has been limited. Our data are consistent with the results of a 2010 study of 49 healthy adolescents by Buralkina N.A., Uvarova E.V. [15], who reported AMH threshold values ​​of 5.24 (0.74) ng/ml. Our findings suggest that the most promising was the use of threshold values of AMH level>7.20 ng/ml and FAI > 2.75, which had the greatest sensitivity and specificity (> 75.0% and 89.0–93.0%).

One of the generally accepted diagnostic criteria for PCOS is polycystic ovaries on ultrasound assessment, which is debatable in adolescence [11]. Moreover, a 2012–2014 study by Grigorenko Yu.P. et al. demonstrated that the use of trans-rectal ultrasound in girls allows better visualization of the ovaries, their volume and follicles than transabdominal ultrasound and improves the diagnosis of the disease [16]. A study of 83 PCOS patients aged 14–16 years showed that girls with PCOS were three times more likely to have the dilatation of the arcuate veins in the uterine plexus and 2.1 times more often had hypervascularization of ovarian stroma than “healthy” girls. Taking into account the size and location of follicles, the authors identified in adolescents with PCOS the same-caliber (57%) and different-caliber (43%) types of ovarian structures, which were associated with the difference in hormonal status and organ size. The authors suggested that using trans-rectal access and 3D visualization the ovaries has several advantages in the diagnosis of PCOS in adolescents. However, this approach requires expert class equipment and is not always readily available screening method. In our study, we used 2D ultrasound and demonstrated that in patients aged 15–17 the diagnostic criterion for PCOS can be the mean ovarian volume> 10.70 cm3 and OUI > 3.95 with a high sensitivity of 81.0–83.0% and specificity of 83.0%. These parameters can be recommended for screening adolescents to identify high-risk patients for further examination.

Conclusion

A girl has an 8.9-fold and 4.9-fold greater chance of developing PCOS if she has a history of neurological disorders during neonatal life and infancy (p = 0.0391) and if her mother had a history gynecological diseases (p = 0.0157), respectively. Mothers of adolescent girls with PCOS with metabolic disorders were more likely to have a history of labor and delivery complications than mothers of girls with PCOS without metabolic disorders and healthy girls (p = 0.0220). Among the diagnostic criteria estimated in adolescents aged 15to17 by 2D pelvic ultrasound, the most sensitive (81.0–83.0%) and specific (83.0%) parameters were the mean ovarian volume>10,70 cm3 and OUI>3.95. The most significant hormonal criteria for PCOS at the age of 15–17 were AMH>7.20 ng/ml, FAI>2.75 ng/ml (sensitivity> 75.0% and specificity 89.0–93.0%), and also T> 1.15 nmol/L, androstenedione> 11.45 ng/ml, and the LH/FSH ratio>1.23 (sensitivity 63.2–78.2% and specificity 84.4–93.7%). Compared with healthy girls, adolescents with PCOS are characterized by a higher BMI [OR 1.31 (95% CI 1.10–1.56), p = 0.0021] and waist circumference [OR 1.11 (95% CI 1,01–1.19), p = 0.0237], as well as a higher cardio-metabolic risk estimated by VAI (p = 0.0461).