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

Disclosures

Clin Endocrinol. 2020;91(1):46-54. 

In This Article

Results

Among 601 enrolled patients, 368 of them (61%) were classified as having CPP. Sixty-three girls with CPP underwent brain MRI which 41 of them (65%) had unremarkable studies. Pituitary microadenoma and Rathke cleft cyst were noted in 16 and 3 patients, respectively. The remaining 3 patients with abnormal brain MRI had middle cranial fossa arachnoid cyst, hippocampal cyst and plagiocephaly. All patients with abnormal MRI findings had no abnormal neurological signs and symptoms. All PT girls had breast regression and no height acceleration within 3–9 months after the initial evaluation.

Clinical characteristics and GnRHa test results of all patients are shown in Table 1. Comparing between CPP and PT girls, weight SDS, height SDS, bone age, bone age advancement and height SDS over MPH SDS were greater in CPP girls. As expected, gonadotrophin profiles and serum oestradiol were higher in the CPP group. Subgroup analysis revealed 127, 153, 241 and 80 girls with CPPII, PTII, CPPIII and PTIII, respectively. Comparisons between CPPII and PTII, and CPPIII and PTIII were performed (Table 1). Despite having comparable MPH SDS, greater height acceleration (height SDS−MPH SDS) was noted in CPPIII as compared with PTIII (1.4 [0.7, 2.1] vs 0.7 [−0.1, 1.5], P < .001), while there was no difference in height acceleration between CPPII and PTII. There were significant differences in bone age, bone age advancement, and hormonal profiles at baseline, peak LH and peak LH:FSH following GnRHa administration between CPPII and PTII, and CPPIII and PTIII. Comparisons between CPPII and CPPIII, and PTII and PTIII are shown in the footnote of Table 1. Peak LH levels following GnRHa test were achieved at 60 minutes in CPP and PT girls, and the serum LH levels remained elevated through 120 minutes (Figure 1). In each Tanner stage, serum LH levels at 60, 90 and 120 minutes after GnRHa administration were not different (data not shown). Considering Tanner stages, greater peak LH level was obtained in CPP girls with Tanner stage III (CPPIII: 9.85 [6.23, 17.61] vs CPPII: 6.96 [4.75, 10.10] IU/L, P < .001), while peak LH levels of PT girls were not different between Tanner stages II and III.

Figure 1.

Serum LH levels during GnRHa testing in (A) all CPP (N = 368) and PT (N = 233) girls, (B) CPP (N = 127) and PT (N = 153) girls with Tanner stage II breasts and (C) CPP (N = 241) and PT (N = 80) girls with Tanner stage III breasts. Error bars show median (horizontal line) and interquartile range ( bar); — CPP; - - - PT; LH, luteinising hormone; GnRHa, gonadotrophin-releasing hormone analogue; CPP, central precocious puberty; PT, premature thelarche

The optimal gonadotrophin cut-offs for differentiating CPP from PT were determined using ROC curves (Figure 2). In girls with Tanner stage II breasts, the AUCs (95% CIs) were 0.887 (0.847–0.926) for peak LH, 0.860 (0.816–0.903) for peak LH:FSH and 0.763 (0.704–0.821) for basal LH. The AUCs (95% CIs) of these 3 parameters in girls with Tanner stage III breasts were greater than those with Tanner stage II breasts (peak LH, 0.953 [0.932–0.974], peak LH:FSH, 0.945 [0.922–0.969] and basal LH, 0.854 [0.813–0.895]). In contrast, basal and peak FSH, basal LH:FSH and oestradiol generated lower AUCs (data not shown). There was wide overlap with basal and peak FSH, and oestradiol between the CPP and PT groups; therefore, these parameters were not useful in diagnosing CPP.

Figure 2.

The ROC curves of hormonal profiles for diagnosing CPP in girls with Tanner stages II (A) and III (B) breasts. ROC, receiver operating characteristic; CPP, central precocious puberty; LH, luteinising hormone; FSH, follicle-stimulating hormone

The diagnostic accuracy of basal and peak LH cut-offs was subsequently determined according to breast Tanner stages (Table 2). This study demonstrated that the basal LH levels of 0.20 and 0.21 IU/L represented the 95th percentile for girls with PTII and PTIII, respectively, while the basal LH level of <0.09 IU/L represented the 5th percentile for both girls with CPPII and CPPIII. Using the basal LH cut-off of 0.2 instead of the traditional proposed cut-off of 0.3 IU/L[1,6,18,19] led to slightly reduced, but acceptable specificity from 98.0% to 94.8% (Tanner stage II) and 98.8 to 93.8% (Tanner stage III). However, sensitivity was increased from 28.3% to 41.7% (Tanner stage II) and 45.2% to 59.3% (Tanner stage III). Interestingly, specificity of basal LH cut-off of 0.2 IU/L was comparable to that of the peak LH cut-off of 5 IU/L in both Tanner stages II (basal: 94.8 vs peak: 90.8%) and III breasts (basal: 93.8 vs peak: 92.5%). Regarding peak LH:FSH, the ratio of 0.43 or greater demonstrated comparable specificity to peak LH cut-off of 5 IU/L (Tanner II: 94.1% vs 90.8%, Tanner III: 92.5% vs 92.5%), but lower sensitivity (Tanner II: 60.6% vs 72.4%, Tanner III: 80.5% vs 85.5%). The traditional proposed peak LH:FSH cut-off of 0.66[20] showed high specificity but low sensitivity in our study.

Peak LH levels in both Tanner stages were positively correlated with basal LH levels (Tanner II: r = .593, P < .001; Tanner III: r = .700, P < .001) and peak LH:FSH (Tanner II: r = .875, P < .001; Tanner III: r = .920, P < .001). A positive correlation between basal LH and basal FSH levels was observed (Tanner II: r = .692, P < .001; Tanner III: r = .773, P < .001).

Figure 3 illustrates a decision tree model using the particular basal and peak LH cut-offs and the probabilities of each clinical event. Peak LH level of 5 IU/L or greater was defined as positivity for GnRHa test, and basal LH levels of 0.2 and 0.3 IU/L were used as the cut-offs. Starting with using basal LH cut-off of 0.2 IU/L, followed by using peak LH cut-off of 5 IU/L in patients with basal LH of <0.2 IU/L in diagnosing CPP, had lower diagnostic cost with more effectiveness, as compared with using basal LH cut-off of 0.3 IU/L, followed by using peak LH cut-off of 5 IU/L in patients with basal LH of <0.3 IU/L, in both Tanner stages II and III (Tanner II: cost 84 vs 91 USD, effectiveness 0.35 vs 0.33; Tanner III: cost 61 vs 72 USD, effectiveness 0.67 vs 0.66). Using basal LH cut-off of 0.2 IU/L followed by peak LH cut-off of 5 IU/L also showed lower diagnostic cost with comparable effectiveness as compared with performing GnRHa test and using a peak LH cut-off of 5 IU/L in all patients (Tanner II: 84 vs 95 USD, 0.35 vs 0.33; Tanner III: 61 vs 95 USD, 0.67 vs 0.64). Therefore, using basal LH cut-off of 0.2 IU/L followed by peak LH cut-off of 5 IU/L in patients with basal LH of <0.2 IU/L provided the most effectiveness at the lowest expense and was considered cost-saving. Comparing between using basal LH cut-offs of 0.2 and 0.3 IU/L followed by performing GnRHa test and using peak LH cut-off of 5 IU/L, the ICERs were 348 USD/detected case in girls with Tanner stage II breasts and 892 USD/detected case in girls with Tanner stage III breasts. The ICERs between a combination of LH cut-offs (basal of 0.2 IU/L followed by GnRHa test using peak of 5 IU/L) and only GnRHa test using a peak LH cut-off of 5 IU/L were 526 USD/detected case in girls with Tanner stage II breasts and 1109 USD/detected case in girls with Tanner stage III breasts.

Figure 3.

Decision tree of peak LH cut-off of 5 IU/L after GnRHa test, basal LH cut-off of 0.2 IU/L followed by peak LH cut-off of 5 IU/L after GnRHa test and basal LH cut-off of 0.3 IU/L followed by peak LH cut-off of 5 IU/L after GnRHa test for diagnosing CPP and PT in girls with Tanner stages II and III breasts. LH, luteinising hormone; GnRHa, gonadotrophin-releasing hormone analogue; CPP, central precocious puberty; PT, premature thelarche. Basal or peak LH levels were defined as positive when they were equal to or greater than the selected cut-offs (basal LH 0.2 and 0.3 IU/L, peak LH 5 IU/L), the probabilities of clinical events in girls with Tanner stages II and III breasts were demonstrated in numbers in solid and dash boxes above and below the horizontal branches, respectively

The results of sensitivity analysis are shown in Figure 4. As compared with either using basal LH cut-off of 0.3 IU/L followed by GnRHa test in patients with negative basal LH or performing GnRHa test in all patients, using basal LH cut-off of 0.2 IU/L followed by GnRHa test remained cost-saving after adjusting for variation of each parameter in both Tanner stages. Cost-effectiveness estimate was most sensitive to the model adjustment for frequency of CPP.

Figure 4.

Tornado diagrams of ICER estimates of basal LH cut-off of 0.2 IU/L followed by peak LH cut-off of 5 IU/L after GnRHa test vs peak LH cut-off of 5 IU/L after GnRHa test in girls with Tanner stages (A) II and (B) III breasts, and basal LH cut-off of 0.2 IU/L followed by peak LH cut-off of 5 IU/L after GnRHa test vs basal LH cut-off of 0.3 IU/L followed by peak LH cut-off of 5 IU/L after GnRHa test in girls with Tanner stages (C) II and (D) III breasts. ICER, incremental cost-effectiveness ratio; LH, luteinising hormone; GnRHa, gonadotrophin-releasing hormone analogue; CPP, central precocious puberty; EV, expected value; USD, US dollar

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....