We performed a retrospective cohort study at a tertiary university affiliated medical center. Included were patients diagnosed with CPP between 1989 and 2011. Data was collected through a review of patient medical records at time of CPP diagnosis and during follow-up visits at the pediatric endocrinology clinic. All patients included in the study presented with thelarche, pubarche (TANNER score of at least 2 prior to the age of 8 years), or bone age advancement, in addition to an early increase in gonadotropin or alternatively with an early menarche. Diagnosis of CPP was based on evaluation of baseline gonadotropins levels (LH). Biochemical diagnostic criteria for CPP included a serum LH concentration of 5 U/L after GnRH (or leuprolide) administration or a basal LH level of 0.3 U/L using ultra-sensitive assays .
Excluded from the study were patients with baseline CNS insults (irradiation, CNS infection or trauma etc.) or pathologies and known endocrinopathies at CPP diagnosis.
We also excluded patients with missing data regarding treatment for CPP or related endocrine and gynecologic disorders at follow-up after 15 years of age. Data collected from medical records included body mass index (BMI) by age, parents' height (measured with a standard wall-mounted stadiometer by the same pediatric endocrinologist (D.G)), past medical treatment or surgeries and presentation at diagnosis. For the pharmacologically treated group of patients, additional information was collected including age at initiation and cessation of treatment, type and duration of treatment, LH levels at treatment initiation and time (months) from cessation of treatment to menarche.
Information regarding follow-up at the age of 15 years or older was collected from the medical records of pediatric endocrinology, general endocrinology and gynecology clinics. Height was measured with accuracy of up to 1 cm and weight of up to 1 kg, BMI (kg /m2) is presented as percentile according to age. Androgen hormones levels – dehydroepiandrosterone-sulfate (DHEA-S), androstenedione and total testosterone, were tested according to the normal reference range reported by the laboratory. Laboratory hyper-androgenism was defined by an increase above the upper normal limit in at least one of these hormones.
Clinical hyperandrogenism was defined by documentation of moderate-severe acne, androgenic hirsutism (defined by a Ferriman-Gallwey score above 8) or hair loss (male pattern). Oligo-menorrhea was defined according to menstrual cycle duration of over 35 days. Anovulation was considered in patients with menses more than 60–90 days apart, at least 2 years after menarche. Menorrhagia was defined as total blood loss exceeding 80 mL per cycle or menses longer than 7 days and metrorrhagia as uterine bleeding at irregular intervals, between the expected menstrual periods. Amenorrhea was defined as lack of menarche by 15 years of age or by 3 years after the onset of breast development (primary) or as lack of menses for more than 3 months or 90 days (secondary).
Polycystic ovary definition was according to the sonographic (by either abdominal or vaginal ultrasound) findings in clinical surveillance and in accordance with the practiced criteria . Polycystic ovary syndrome was diagnosed according to the Androgen Excess Society 2006 definition  that required the following two parameters: (1) Clinical or laboratory hyperandrogenism and (2) Oligo-menorrhea or sonographic evidence of polycystic ovaries. When evaluating adolescents for PCOS, we used the Diagnostic criteria for polycystic ovary syndrome in adolescents presented by Carmina et al.  as these criteria were most commonly used in our institutions' practice.
For all patients in the study data including parameters relevant to PCOS  were available for at least 2 years after menarche. In addition, data on the presence of endocrine (including hyperinsulinism, defined as a fasting plasma insulin level higher than 2 μU/mL), metabolic or gynecologic pathologies, as well as obstetric history documented at the follow up, were collected.
The study was approved by the institutional ethical review board (IRB number HMO -192–19).
To test the relationship between categorical variables, the Chi square test or Fisher's exact test were used. Quantitative variables were compared between 2 independent groups using the Student's t-test for normally distributed variables or the Mann–Whitney U test for quantitative variables not normally distributed.
Univariate analysis was performed for factors associated with PCOS in adulthood in for the entire study population and included age, BMI percentile, clinical signs of puberty at CPP diagnosis, LH, FSH and their ratio at CPP diagnosis and treatment with GnRHa. We report odds ratios (OR), 95% confidence interval (CI) for parameters included in the final analysis.
All statistical tests were two-tailed, with a p-value of 0.05 or less considered statistically significant. All statistical calculations were performed using the statistical software package SPSS 24.0 (SPSS Inc., Chicago, IL).