Study design and participant selection
This was a retrospective case-control study. The cases consisted of IPT girls with significantly advanced BA (BA minus CA (BA-CA) ≥2 years). The controls were IPT girls without significantly advanced BA (BA-CA <2 years). Patients admitted to the Endocrinology Department of Shenzhen Children’s Hospital from January 2015 to April 2018 who needed a further gonadotropin-releasing hormone (GnRH) stimulation test due to suspected precocious puberty were enrolled as the study subjects.
Most of girls aged 4-8 years with breast development were followed up in our outpatient clinic. They were potentially hospitalized for further investigations when they presented with no breast tissue regression for at least three months and one or more of the following: 1) Progressive breast development: the progression of breast Tanner stage 1 was less than half a year; 2) Significant BA advancement: BA was greater than 2 years above CA; 3) Linear growth acceleration: the growth rate increased by over 1.5cm/ year compared with the previous growth rate.
The inclusion criteria were as follows: 1) Age at onset of the disease was 4-8 years; 2) Patients with isolated breast development; 3) Patients without other secondary sexual characteristics, pigmentation of the nipple and areola, and growth of pubic hair or armpit hair; 4) Peak luteinizing hormone (LH) level was less than 5 IU/L and the ratio of peak LH to peak follicle-stimulating hormone (FSH) in the GnRH stimulation test was less than 0.6; 5) Breast development had no other causes, such as peripheral precocious puberty, exogenous estrogen intake, local mammary gland hyperplasia and so on. There had been 480 IPT girls included in the study initially.
Children with history of hormone treatment, history of intaking nutritional tonic and Chinese traditional medicine, other endocrine disease (including growth hormone (GH) deficiency, thyroid dysfunction, adrenocortical dysfunction or hyperactivity), abnormal bone development, congenital dysplasia, premature birth and small for gestational age were excluded. Ultimately, a total of 423 participants were enrolled in the study (Fig. 1).
According to the values of BA-CA, the participants were divided into the advanced BA group (BA-CA ≥2 years) and the control group (BA-CA <2 years). According to body mass index (BMI)[8], the participants were classified into three groups: normal weight (BMI P5 ~ P85 or BMI standard deviation score (SDS) -1.65 to <1.04), overweight (BMI P85 ~ P95 or BMI SDS 1.04 to <1.65) and obesity (BMI≥P95 or BMI SDS ≥1.65). IPT girls with normal weight were then subdivided into two groups separately according to the levels of insulin-like growth factor-1 (IGF-1) SDS and dehydroepiandrosterone sulfate (DHEAS) SDS.
Height and weight were measured, and BMI was calculated as body weight (kg)/height (m2). These values were expressed as age- and sex-specific SDS based on national survey data in 2005[8, 9]. Breast development was staged according to the Tanner criteria. If the Tanner stages of bilateral breast development were different, the mature side was recorded. Hypothalamic-pituitary-gonadal axis function was evaluated by basal sexual hormone levels and the GnRH stimulation test (gonadorelin injections at 2.5-3.0 μg/kg, with no more than 100 μg administered at a time). Other hormones, including oestrogen, DHEAS, androsterone, sex hormone binding globulin (SHBG), IGF-1 and insulin, were evaluated by enzyme-enhanced chemiluminescence assays (Siemens IMMULITE 2000, Munich, Germany). IGF-1 and DHEAS levels were standardized according to reference data for different ages in China[10, 11]. Lipids, including triglycerides, total cholesterol, high-density lipoprotein and low-density lipoprotein, were detected by routine biochemical assays.
An X-ray of the hand and wrist was taken, and the AI BA evaluation technology was developed by Yitu Healthcare (Hangzhou, China). This product is version 3 of the Intelligent Diagnosis System for Child Growth and Development, which utilized RUS-CHN. All participants underwent a pelvic ultrasound examination (GE Logic E9 ultrasound machine with a 3.5-12 MHz convex probe and a 9-12 MHz linear probe).
Ethical approval
This study was approved by the Ethics Committee of Shenzhen Children’s Hospital(No. N2019056).
Statistical analysis
The t-test was used to compare advanced BA group and control subjects for the following continuous measures: height (SDS), weight (SDS), BMI (SDS), and BA-CA. Data regarding age at onset, CA, disease duration, uterine length, ovarian volume, basic and peak LH, Peak LH/peak FSH, DHEAS (SDS), IGF-1 (SDS), androsterone, insulin, SHBG, oestrogen and lipids were clearly skewed, so median and quartiles were used for these numerical variables. Those differences between two groups were compared using a Mann-Whitney U test. The association between significant BA advancement and breast Tanner stage as well as status of obesity or overweight were calculated using the chi-square test. Comparison of IGF-1 SDS and DHEAS SDS with BA advancement in normal weight subgroup were analyzed by the t-test.
In order to assess the independent risk factors affecting significantly advanced BA, we performed a binary logistic regression analysis (forward LR stepwise method) with the advanced BA of more than 2 years above CA (yes/no) as the dependent variable. The independent variables were those shown to be significant (P<0.05) in the comparison between two groups (disease duration, BMI SDS, DHEAS SDS, IGF-1 SDS, androsterone, insulin, SHBG, uterine length), significant factors in previous studies (oestrogen), and other factors that may be clinically correlated (basic LH, peak LH, ratio of peak LH to peak FSH, ovarian volume).
A multiple linear regression model was then used to predict the factors most likely to advance BA, with BA-CA as the dependent variable. After testing the assumptions of the model, we entered the factors shown to be significant (P<0.05) in the comparison between two groups (BMI SDS, DHEAS SDS, IGF-1 SDS, insulin, SHBG, uterine length) and others that may be clinically correlated (peak LH, ratio of peak LH to peak FSH, ovarian volume) as the independent variables.
Statistical analyses were performed using SPSS version 23.0. A p value <0.05 was considered to indicate statistical significance.