Precocious puberty is generally defined as the onset of secondary sexual characteristics before 8 yr of age in girls and 9 yr in boys [1, 2]. Precocious puberty caused by central activation of the hypothalamic-pituitary-gonadal (HPG) axis is classified as gonadotropin-dependent, also called true or central precocious puberty (CPP). Isolated forms of abnormal pubertal development, such as premature thelarche (PT), are not associated with central activation of the HPG axis, and are classified as peripheral precocious puberty [3]. CPP occurs at least 10-fold more frequently in girls than in boys; in more than 90% of girls, sexual precocity is idiopathic. It is estimated that precocious puberty affects between 1 in 5,000 and 1 in 10,000 children, and a recent study found a much higher incidence of CPP in girls compared with 40 yr ago [4, 5]. Gonadotropin releasing hormone (GnRH) stimulation test is the gold standard to reveal premature activation of the HPG axis in patients with precocious puberty [6, 7].

However, this test needs a clinical investigation unit and repeated blood sampling at 5-8 time points to measure the levels of luteinizing hormone (LH) and follicle-stimulating hormone (FSH) [8-10]. The duration of the test is 90 to 120 min and 15 to 25 mL of blood is required. As such, this test is comparatively expensive, time-consuming, painful, and uncomfortable for patients [9-13]. To avoid these problems, several alternatives such as measurement of basal LH and FSH values, and the subcutaneous leuprolide acetate test with a single sample, have been attempted [14-17]. None of these alternative methods have yet been standardized sufficiently or proven equal or superior to GnRH test.

In this study, we investigated whether the GnRH test to diagnose CPP could be simplified without altering its validity, by estimating the values of LH and FSH at various time points after GnRH stimulation.

A hundred and sixty-six girls showing signs of early puberty were enrolled in this study. In 128 out of 166 (77.1%) tests, the peak LH was above the cut-off value, and the test was diagnostic for CPP. The remaining girls (38 out of 166, 22.9%) had an evoked LH value of <5 IU/L, and were classified as undergoing PT. The clinical characteristics and hormone levels of both groups are presented in Table 1. Significant differences in chronological age and bone age were seen between the 2 groups. Basal LH and FSH levels in the group with CPP were higher than in the PT group (P<0.001). Evaluation of the LH/FSH ratio demonstrated statistically significant differences between the CPP and PT groups in both the basal and peak states (P<0.001).

Basal and peak LH levels in the CPP group were significantly greater than in the PT group (both, P<0.001) (Fig. 1). Both LH and FSH levels showed a rapid increase after GnRH stimulation in both groups. However, the increase in LH levels was more robust in the CPP group than in the PT group, while FSH responses were almost similar in the 2 groups (Fig. 2). The peak LH level was achieved 30 min after GnRH stimulation in the CPP group, and 45 min after stimulation in the PT group, while FSH levels continued to rise throughout the test in both groups (Fig. 2). The peak LH/FSH ratio was achieved in 15 min after GnRH stimulation in both groups (Fig. 2).

Among CPP patients, 98.4% (126 of 128 tests) of the 45-min samples were diagnostic for CPP, with LH levels of ≥5 IU/L (Table 2). The 2 patients who had stimulated LH levels below this cut-off value at 45 min, showed values above the cut-off at 30 min. Therefore, the cumulative frequency of LH values of ≥5 IU/L at 45 min was 100% (Table 2).

Fig. 3 shows the ROC curves of LH at different time points of the test in all patients. The ROC curve for LH at 45 min had the greatest area under the curve (AUC) and was hence the most effective for diagnosing CPP. 5 IU/L was considered as the cut-off value for LH in the diagnosis of CPP, and the ROC curves indicate that an LH cut-off value of 4.99 IU/L at 45 min resulted in high sensitivity (98.4%) and specificity (100%) in the diagnosis of CPP.

Pubertal development of breast was compatible with Tanner stages 2 and 3 in 89.8% and 10.2% of girls, respectively. Pubic hair development was compatible with Tanner stages 1 and 2 in 95.2% and 4.8% of girls, respectively.

CPP may have profound physical and psychological effects on affected children and their families. It is accompanied by growth acceleration, advancement of bone age, and elevated sex steroid hormone levels for age. Therefore, early menarche and significant impairment of final height can result in untreated CPP patients. Mogensen et al. [20] reported a significant increase in the number of CPP patients over the recent 16-yr period. This trend also occurred in Korean girls with a 4- to 5-fold increment in CPP patients over the recent 5-yr period.

The gold standard for laboratory confirmation of CPP is the GnRH stimulation test. However, this test results in significant cost, time, and discomfort to the patient, as well as a potential delay in diagnosis. Different cut-off levels of peak LH are used in the diagnosis of CPP in patients with early pubertal signs. Further, cut-off levels differ depending on the assays used to measure LH [19, 21, 22]. Previous studies have suggested that GnRH-stimulated LH levels between 30 and 60 min are sufficient to diagnose CPP [21, 23-25]. In Korean patients with CPP, Choi et al. [26] tried to simplify the standard GnRH stimulation test, suggesting that a single LH determination at 30 min was diagnostic. However, their study included only 33 patients. We analyzed the LH and FSH response to the GnRH test performed to diagnose precocious puberty in 166 patients. To our best knowledge, the current study comprises the largest homogeneous Korean population in this regard.

We measured LH levels by a direct chemiluminescence immunoassay, and a cut-off stimulated LH value of ≥5 IU/L was considered diagnostic for CPP. In 126 out of 128 (98.4%) CPP patients tested, 45-min samples were diagnostic for CPP using this cut-off value. The remaining 2 patients, who had stimulated LH levels of <5 IU/L at 45 min, had levels above the cut-off value at 30 min. Therefore, the cumulative frequency of LH values of ≥5 IU/L at 45 min was 100%. These findings indicate that 2 samples, obtained at 30 and 45 min after stimulation, can be used in a simplified GnRH test for discriminating patients with CPP from patients with PT.

Neely et al. [14] reported that a basal LH value of >0.1 IU/L was diagnostic for CPP with 94% sensitivity and 88% specificity. They also reported that a cut-off value of >0.3 IU/L increased the specificity to 100%, although the sensitivity decreased. In our study, the basal LH level in the group with CPP was higher than that of the PT group (0.42 and 0.07 IU/L, respectively, P<0.001). However, analysis of the diagnostic value of basal LH using ROC curves revealed low sensitivity and specificity of basal LH levels for the diagnosis of CPP (data not shown). In addition, we observed that 29 of the 128 patients (22.7%) diagnosed with CPP by the GnRH stimulation test showed undetectable basal LH levels (LH <0.07 IU/L). Therefore, an elevated basal LH was highly predictive of a positive GnRH test result, whereas a low basal LH did not eliminate central pubertal activation. We, therefore, do not suggest using basal LH as a single diagnostic marker for CPP.

The LH/FSH ratio has been considered of value by some investigators who reported that CPP in girls is usually accompanied by an LH/FSH ratio of >1.0 [8]. However, nocturnal LH secretion and the LH response to GnRH may be low at early puberty, and the LH/FSH ratio may remain low until mid-puberty [22]. Changes in this ratio at puberty are predominantly caused by increasing LH levels. In the current study, LH/FSH ratio showed significant differences between CPP and PT groups in both basal and peak states (P<0.001), with the basal and peak LH/FSH ratios being 0.14 and 0.74, respectively, in the CPP group. A basal LH/FSH ratio of >0.2 as a diagnostic tool for CPP had a sensitivity of 36.7% and specificity of 97.4%. With an LH/FSH ratio cut-off value of >0.5, the diagnostic sensitivities at 15, 30, 45, 60, 90, and 120 min were 77.3%, 76.6%, 69.5%, 57.8%, 40.1%, and 35.2%, respectively, while the specificities were 94.7%, 97.4%, 98.9%, 100%, 100%, and 100%, respectively. The limited utility of the LH/FSH ratio seems to be a consequence of the considerable variation of FSH levels seen in patients.

Although FSH concentrations were also observed to increase in response to the GnRH stimulation test in this study, FSH levels were indistinguishable between the CPP and PT groups, and the peak FSH levels of the CPP and PT groups had poor diagnostic value (P=0.748). Data from other studies also indicate that FSH has poor diagnostic value for CPP [10, 14, 21]. Therefore, FSH concentrations cannot be used in diagnosing CPP, and the accurate measurement of LH suffices for diagnostic confirmation of CPP.

In conclusion, a single LH value measured 45 min after GnRH stimulation is highly sensitive and specific for the diagnosis of CPP, using a cut-off value of ≥5 IU/L. Two samples, obtained at 30 and 45 min after stimulation, were able to accurately diagnose all CPP patients in this study. Although further study is required, we suggest that repeated blood sampling is unnecessary, and that 2 samples obtained at 30 and 45 min can be used in a simplified GnRH test to evaluate precocious puberty. The proposed simplified test will substantially diminish the cost and discomfort to the patient.