Basal Serum Luteinising Hormone Cut-Off, and Its Utility and Cost-Effectiveness for Aiding the Diagnosis of the Onset of Puberty in Girls With Early Stages of Breast Development

Somboon Wankanit; Pat Mahachoklertwattana; Oraluck Pattanaprateep; Preamrudee Poomthavorn


Clin Endocrinol. 2020;91(1):46-54. 

In This Article

Materials and Methods


This retrospective study was conducted at Pediatric Endocrine Clinic, Ramathibodi Hospital (a tertiary hospital), Mahidol University, Bangkok, Thailand. Medical records of girls who had breast development before 8 years of age and underwent subcutaneous GnRHa test for CPP diagnosis between October 2007 and February 2019 were reviewed. Only Tanner stages II and III breasts were documented in all girls. Girls who were diagnosed as having benign PT before 3 years of age or gonadotrophin-independent precocious puberty and those without follow-up clinical course were excluded. All enrolled girls had neither chronic illness nor medication use. Clinical characteristics including weight, height, body mass index (BMI) and breast stage were recorded. Standard deviation scores (SDSs) of weight and height were calculated using the National Standard Growth Curve of the Ministry of Public Health, Thailand.[11] BMI SDS was calculated using World Health Organization standards.[12] Mid-parental height (MPH) in centimetres was derived from the formula: (maternal height+paternal height−13)/2, and its SDS was calculated. Breast stage was determined using Marshall and Tanner classification.[13] Additionally, bone age X-ray results according to Greulich and Pyle standard were collected.[14] Brain magnetic resonance imaging (MRI) results were retrieved if available.

Pubertal progression in all patients was observed without depot GnRHa treatment during a 3- to 6-month period of follow-up after subcutaneous GnRHa testing. During that period, girls who had progressive breast development concomitant with accelerated height velocity, advanced bone age at the initial presentation and pubertal-sized uterus and ovaries were diagnosed as having CPP. Girls who had no aforementioned findings were classified as having PT. In each group, patients were further divided into two subgroups according to their breast Tanner stages (II and III) at the time of GnRHa testing. Therefore, there were four groups of patients; CPP girls with breast Tanner stages II and III (CPPII and CPPIII) and PT girls with breast Tanner stages II and III (PTII and PTIII). The GnRHa test results were analysed and compared among these four groups.

Subcutaneous GnRHa Testing

Subcutaneous GnRHa testing was performed in the morning according to the standard protocol of our institute.[4] Following insertion of intravenous cannula for blood sampling, blood samples were collected for measurements of basal serum LH, follicle-stimulating hormone (FSH) and oestradiol. Triptorelin (Diphereline®, Ipsen Pharma Biotech) at a dose of 100 micrograms was then administered subcutaneously. Serum LH was measured at 60, 90 and 120 minutes after the triptorelin injection. Peak serum LH was defined as the maximum level achieved during the testing. Peak serum FSH was determined at 120 minutes after the triptorelin injection. Chemiluminescent microparticle immunoassay (CMIA) by Architect i2000SR analyzer® (Abbott), which was demonstrated to have consistent performance when compared with the other widely used immunoassays (such as immunoluminescence assay of Liaison®, Diasorin; electrochemiluminescence assay of Elecsys®, Roche Diagnostics GmbH; and immunoradiometric assay of Adaltis®),[15] was used for analysing serum LH and FSH levels. The lower limits of detection for serum LH and FSH levels were 0.09 and 0.05 IU/L, respectively. The intra-assay and interassay coefficients of variation (CV) were 1.9%-3.6% and 2.4%-3.9% for serum LH, respectively, and 2.6%-4.2% and 3.2%-4.6% for serum FSH, respectively. The intra-assay CV for the low LH values (0.09–0.3 IU/L) was 7.2%. Serum oestradiol concentration was analysed by electrochemiluminescence assay, using Cobas e 602 analyzer® (Roche). The lower limit of detection was 18.4 pmol/L with intra-assay CV of 1.1%-6.7% and interassay CV of 1.9%-10.6%.


The study was approved by the Ethics Committee of the Faculty of Medicine Ramathibodi Hospital, Mahidol University (Date 21 November 2017, MURA 2017/776) and conformed to the Declaration of Helsinki.

Statistical Analysis

The data were analysed using SPSS Statistics version 24.0 (SPSS Inc) and Stata version 15.1. Owing to non-normally distributed data, they were presented as median (interquartile range, IQR). Mann-Whitney U test was applied to evaluate the differences of continuous data between two groups of patients. Correlation between variables was performed using Spearman's analysis. The receiver operating characteristic (ROC) curves of GnRHa test results were constructed to assess area under the curve (AUC) and 95% confidence interval (95% CI). The sensitivity and specificity of serum LH cut-offs for diagnosing CPP were then determined. Statistical significance was considered when P value was <.05.

Cost-effectiveness Analysis

A decision tree model was constructed using TreeAge Pro® 2016 R2 release (TreeAge Software Inc) to compare the cost-effectiveness of different basal LH cut-offs. The basal LH cut-offs which provided comparable specificity to the selected peak LH cut-off were assessed to minimize misdiagnosis of CPP. Whether performing GnRHa testing only in girls who had basal LH level below the particular cut-offs would be more cost-effective than performing GnRHa testing in all girls with suspected CPP was also determined. Owing to the limitation of this retrospective study, only hospital perspective was analysed. Direct medical costs were diagnostic costs based on data of our hospital in the year 2019, including basal serum LH measurement (9.4 USD) and GnRHa testing (94.9 USD), with 1 USD = 31.93 THB.[16] Effectiveness was the proportion of the number of CPP girls who were diagnosed by using our proposed LH cut-offs to the number of patients classified by their diagnoses and levels of basal or peak LH. Therefore, effectiveness was defined as 1 in CPP girls with positive LH levels for the particular cut-offs. Otherwise, it was equal to 0. Incremental cost-effectiveness ratio (ICER) was calculated as follows:

One-way sensitivity analyses were performed and shown in Tornado diagrams to evaluate the model's sensitivity to uncertainty of 3 input parameters in determining the ICER. These parameters included 10% variation in costs of basal LH measurement and GnRHa testing, and frequency of CPP among girls who were evaluated for precocious puberty which varied between 20% in a previous study,[17] and 45% (Tanner II) and 75% (Tanner III) in our study.