All female children with CPP were recruited from The Fifth Affiliated Hospital of Harbin Medical University. The study was approved by the ethics committee of The Fifth Affiliated Hospital of Harbin Medical University (KY [2018] 002). CPP was diagnosed according to the 2015 edition of the Chinese guidelines for PP prevention and treatment. The detailed evaluation criteria are as follows: (1) early development of secondary sexual characteristics: The development of secondary sexual characteristics in girls and boys occurred before the ages of 8 and 9 years, respectively. (2) Linear growth acceleration: The annual growth rate was higher than that in healthy children. (3) Increased bone age: Bone age exceeded the actual age by ≥ 1 year. (4) Gonadal enlargement: Pelvic ultrasonography showed an increase in the female uterus and ovary volume and multiple follicles of a diameter of > 4 mm in the ovary; the testis volume was > 4 ml in boys. (5) The hypothalamic–pituitary–gonadal axis (HPGA) function was initiated, and the serum gonadotropin and sex hormones reached pubertal levels.

The healthy control group (group N; 30 female children) was formed based on the physical examination of female children. The disease group (group S; 31 female children) included female children with CPP. The initial diagnosis of the sera sample of the 31 female children was collected as a disease group. The treatment group was based on our available traceable cases (group R; 29 female children). A chemical drug (a GnRH analog) and a traditional Chinese medicinal formula were combined to treat female children with CPP in 6 months (group R). Non-targeted metabolomic analysis was performed to analyze differences in serum metabolites among the healthy control (group N), CPP (group S), and treatment (group R) groups. The GnRH analog used was injectable leuprorelin acetate under the trade name Bey [National Medical Products Administration (NMPA) ratification number: Guo Yao Zhun Zi-H20093852; lot number of the existing pharmaceutical administration: 210614]. The traditional Chinese medicinal formula used was DBYW (NMPA ratification number: Guo Yao Zhun Zi-Z20043458).

Sample information is summarized in Table 1.

N, healthy female children (n = 30); S, female children with central precocious puberty (n = 31); R, female children treated with drugs used in this study (n = 29). All data are expressed as mean ± SE (two-tailed t-test).

The level of luteinizing hormone (LH) and follicle size (left and right) have returned to normal. In addition, the ovary volume decreased; however, it did not return to normal and at least showed a downward trend from the disease group.

Chromatography was performed on an ultra-high-pressure liquid chromatography (UHPLC) system (Agilent 1290, USA). The chromatographic column used was ACQUITY UHPLC HSS T3 (1.8 μm; 2.1 × 100 mm) (Waters). Mobile phase A included 0.1% formic acid in the water, and mobile phase B included 0.1% formic acid in acetonitrile (Zhao et al., 2014). Gradient elution was performed as follows: 5% B for 1 min, which was changed linearly to 10% B within 1 min, which was subsequently changed linearly to 95% B within 12 min and held for 2 min, which was eventually changed linearly to 5% B within 1 min and held for 3 min. The analytical column and autosampler temperatures were 35°C and 4°C, respectively. The sample volume was 5 μl for each run, and the column eluent was directly analyzed on the MS system (Zhao et al., 2014).

For analysis, accurately 200 μl of serum was mixed with 4 volumes of methanol/acetonitrile (1:1, v/v). All samples were shaken for 30 s and subjected to ultrasonication for 10 min in ice-cold water. The samples were incubated at -20°C for 2 h to facilitate the precipitation of protein and were subsequently subjected to centrifugation at 13,000 × g for 15 min at 4°C. The serum supernatant was collected, dried under a vacuum, and subjected to centrifugation at 4°C. Subsequently, the aliquots were resuspended in 200 μl of acetonitrile/water (1:1), shaken for 30 s, and subjected to centrifugation at 13,000 × g for 15 min at 4°C, and the supernatants were collected. Approximately 150 μl of the supernatant was analyzed on a UHPLC system (Agilent 1290, USA) coupled with a Q-TOF system (Agilent 6545, USA). The remaining supernatants were mixed to prepare quality control (QC) samples. QC was performed after every 15 serum samples were analyzed.

Data were obtained using the auto MS/MS mode from 50 to 1,100 m/z. Collision energies for collision-induced dissociation were 20 and 40 V. MS parameters were set as follows: ion source dry gas temperature, 320°C; N₂ gas flow, 8 L/min; sheath gas temperature, 350°C; sheath gas flow, 12 L/min; and ion spray voltage, 4,000 (positive ion) and 3,500 V (negative ion).

Data files from the Q-TOF-MS system were converted to the .abf format using the Analysis Base File Converter software. Peak detection, chromatogram deconvolution, and other data analyses were performed using the MSDIAL3.82 software with the following parameters: alignment-MS1 tolerance, 0.01 Da; retention time tolerance, 0.2 min; identification accurate mass tolerance (MS1), 0.005 Da; MS2, 0.05 Da; and identification cutoff score, 60%. The HMDB,¹ METLIN,² Massbank,³ and KEGG⁴ databases were used to identify key metabolites and metabolic pathways.

Multivariate statistical analysis was performed using the SIMCA-P software (version 14.1; Umetrics, Umea, Sweden). The matrix was imported into R to observe the overall distribution of samples and assess the stability of the whole process using principal component analysis (PCA). Partial least squares discriminant analysis (PLS-DA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were used to distinguish metabolites between groups. To evaluate the quality of the model, sevenfold cross-validation and 200 permutation tests [response permutation testing (RPT)] were performed. The permutation test was used to assess the validity and degree of overfitting of the model. VIP values of > 1.0 and p-values of < 0.05 indicated significant differences. The Mann–Whitney U test was used to compare two groups, with p-values of < 0.05 indicating significant differences. The false discovery rate (FDR) was used for comparing multiple groups (p < 0.10). The ratios of different metabolites between the average of those in the healthy control group and the two experimental groups were calculated. Figure 1A is created using the hiplot online website⁵ and Figure 1B using Wekemo Bioincloud.⁶

To identify crucial metabolites associated with the treatment of CPP, the serum metabolic profiles of female children in the three groups (N, S, and R groups) were analyzed via UHPLC-TOF-MS. Data analyzed using the SIMCA-P software, 3D-PCA, and PLS-DA are demonstrated in Figures 2A, B. OPLS-DA was also used to screen for critical metabolites in each group (Supplementary Figures 1A–C). Differences in metabolites were analyzed between the S and N, R and N, and R and S groups. All OPLS-DA models were reliable because the permutation tests did not reveal overfitting (Supplementary Figures 1D–F). Several critical metabolites and metabolic pathways associated with treatment were identified in the serum of female children (Supplementary Table 1).

Differentially expressed metabolites were analyzed among the N, S, and R groups (Figure 2A and Supplementary Figures 2A, B). A total of 21 metabolites were found to be upregulated in the S group compared with the N group (Figure 3A). In addition, a total of 28 downregulated and 21 upregulated metabolites were identified in the R group (Supplementary Figure 2A). A total of 55 differentially expressed metabolites were identified between the R and N groups (Supplementary Figure 2B), including 24 downregulated and 31 upregulated metabolites. These findings reveal that some specific metabolites play a significant role in the treatment of CPP.

Pearson’s correlation analysis was performed to examine the correlation among the identified metabolites in female children with CPP. Figure 1A shows the correlation among the top 50 differentially expressed metabolites. Lys-Met-His, one of the oligopeptides, had a significant positive correlation with ilicic acid, 7-hydroxy-coumarin, 6-keto PGF1α, and N-cis-hexadec-9Z-enoyl-L-homoserine lactone. Figure 1B shows the network analysis of the 46 identified metabolites in female children with CPP. The identified metabolites were classified as lipids, short peptides, carnitines, glucides, amino acids, nuclein, acids, and others.

A total of 14 pathways (including aminoacyl–tRNA biosynthesis; phenylalanine, tyrosine, and tryptophan biosynthesis; and fatty acid biosynthesis) were significantly enriched in the three groups (Figure 4). As shown in Figure 1A, phenylalanine was positively correlated with Phe-Phe, hypoxanthine, (R)-(+)-2-pyrrolidone-5-carboxylic acid, Lys-Lys-Thr, 12-HETE, and Asn-Leu-Pro-Ala-Lys but was negatively correlated with trans-trans-muconic acid, 6-keto PGF1α, and desalkyl verapamil D617. Tyrosine was positively correlated with 2-aminoacetophenone, p-coumaric acid, indoline, phenylalanine, and isoleucine but was negatively correlated with N-cis-hexadec-9Z-enoyl-L-homoserine lactone, 6-keto PGF1α, and trans-trans-muconic acid.

As shown in Figures 4A, B, aminoacyl–tRNA biosynthesis was the most significantly enriched pathway in the R group. In Figure 4B, the abscissa indicates the rich factor, which was calculated by dividing the number of differentially expressed metabolites in the corresponding metabolic pathway by the number of total metabolites identified in the pathway. The higher the rich factor, the larger the degree of pathway enrichment. Colors are expressed on a green to red linear scale with gradually decreasing p-values. The bigger the bubble, the more the number of metabolites in the pathway.

Short peptides contain fundamental biological information and are a precursor to life. Scholars have recently developed a greater interest in short peptides owing to their unique features and prospects for developing innovative bio-therapies (Apostolopoulos et al., 2021). In this study, two short peptides, namely, His-Arg-Lys-Glu and Lys-Met-His, were differentially expressed between the N and S groups (Figure 5). In addition, eight short peptides, namely, Arg-Trp-Glu, His-Arg-Lys-Glu, Leu-Val-Val-Val-Asp, Lys-Lys-Thr, Lys-Met-His, Ser-Ala-Val, Thr-Val-Leu-Thr-Ser, and Tyr-His-Val, were differentially expressed between the R and S groups. These peptides were upregulated after drug treatment, except for Lys-Met-His and Ser-Ala-Val.

Furthermore, eight short peptides, namely, Arg-Trp-Glu, Asn-Leu-Pro-Ala-Lys, Leu-Val-Val-Val-Asp, Lys-Lys-Thr, Lys-Met-His, Ser-Ala-Val, Thr-Val-Leu-Thr-Ser, and Tyr-His-Val were differentially expressed between the N and R groups. The short peptides His-Arg-Lys-Glu and Lys-Met-His may play a significant role in CPP among female children and may be potential targets for the treatment of CPP. Furthermore, Lys-Met-His was positively correlated with endiline, ilicic acid, 7-hydroxy-coumarin, desalkyl verapamil D617, 6-keto-PGF1α, enterolactone, and N-cis-hexadec-9Z-enoyl-L-homoserine lactone (Figure 1A) but was negatively correlated with Lys-Lys-Thr, isoleucine, Asn-Leu-Pro-Ala-Lys, capsidiol, and myristoleic acid.

As shown in Figure 6, the expression of 5 amino acids, namely, isoleucine, N-fructosyl isoleucine, indoline, phenylalanine, and tyrosine, was lower in the R group than that in the S group. In addition, the expression of L-(+)-gulose, isoleucine, N-fructosyl isoleucine, and tyrosine was lower in the R and N groups. As shown in Figure 6, the drug may act by downregulating the above-mentioned five amino acids, which are closely involved in the modulation of CPP among female children.

The results of the analysis of lipid metabolites are shown in Figure 6. As shown in the figure, the drug may act by regulating PE [19:1(9Z)0:0]. As shown in Figure 1A, PE [19:1(9Z)0:0] had a positive correlation with three metabolites [tumonoic acid I, PC (16:0/0:0), and Phe-Phe] and a negative correlation with one metabolite (trans-trans-muconic acid).

Differentially expressed acid metabolites were identified to distinguish among female children in the three groups. As shown in Figure 7, tumonoic acid I and palmitic amide may be associated with the pathogenesis of CPP. The expression of tumonoic acid I was increased and that of phthalic anhydride was decreased in the CPP group; however, these changes were reversed after drug treatment.

As shown in Figure 1A, tumonoic acid I showed a significant positive correlation with Phe-Phe, 12-HETE, hypoxanthine, and Lys-Lys-Thr but a negative correlation with trans-trans-muconic acid, ilicic acid, and 6-keto PGF1α. In addition, palmitic amide was positively correlated with stearamide, oleic acid, and Lys-Met-His.

It remains unclear whether one or more biomarkers are involved in certain physiological processes or pathological alterations in CPP; however, linoleic acid–biotin can be one of them in Figure 8. In this study, the expression of linoleic acid–biotin was higher in the S group than that in the N group but was decreased after drug treatment. In addition, linoleic acid–biotin was positively correlated with PE [19:1(9Z)/0:0], PC (16:0/0:0), tumonoic acid I, isoleucine, and Lys-Lys-Thr and was negatively correlated with uric acid, ilicic acid, 7-hydroxy-coumarin, desalkyl verapamil D617, 6-keto PGF1α, and N-cis-hexadec-9Z-enoyl-L-homoserine lactone (Figure 1A).

The expression of 12-HETE, hypoxanthine, and capsidiol was different between the S and N groups.

Figures 1, 3, 4–8 of the results provided significant insights into drug therapy mechanisms and personalized treatment, but we also analyzed ROC curves for metabolites. The correlation between the levels of palmitic amide, PE [19:1(9Z)/0:0], and tumonoic acid I and between the treatment response and progression of CPP among children was analyzed (Figure 9 and Supplementary Table 2). The AUC values of the three metabolites were 0.7441, 0.7054, and 0.7312, respectively.

His-Arg-Lys-Glu, Lys-Met-His, PE [19:1(9Z)0:0], tumonoic acid I, palmitic amide, and linoleic acid–biotin may play a significant role in CPP among female children.

Several short peptides (2–20 amino acids) are partially obtained via enzymatic hydrolysis of protein and are important bioactive peptides (Zhang et al., 2019). In recent years, short peptides have been of considerable interest in the branch of biology, chemistry, and medicine owing to their unique structural and functional diversity (Apostolopoulos et al., 2021). Short peptides play an important role in the immune system and are responsible for transmitting most immunological information (Parhiz et al., 2013). Studies have reported that short peptides play an important role in many diseases, including neurodegenerative diseases (Jang et al., 2016), Alzheimer’s disease (Russell-Aulet et al., 1999; Frutos et al., 2007; Tatman et al., 2016; Darawshe et al., 2019), and rheumatoid arthritis (Darawshe et al., 2019), and have therapeutic effects (New et al., 2014; Tatman et al., 2016; Baig et al., 2018; Horsley et al., 2020). In this study, two short peptides were differentially expressed between the N and S groups, whereas eight short peptides were differentially expressed between the R and S groups (Figure 6). These findings warrant further investigation and may serve as a reference for developing new therapeutics for CPP in female children.

Furthermore, the expression of PE [19:1(9Z)0:0] was higher in female children with CPP than in healthy female children but was decreased after drug treatment. PE [19:1(9Z)0:0], a phosphatidylethanolamine, is a brain phospholipid that plays an important role in improving reflex development, cognition, and memory (Müller et al., 2015; Li et al., 2016; Kim et al., 2019). As a breakdown product of PE, lysophosphatidylethanolamine (LPE) is present in the cells of all organisms. Alternatively, the perturbations of LPE may reflect PE dysfunction, which may be related to changes in the structural membrane and permeability of the plasma membrane (Gao et al., 2017).

The specific effects of tumonoic acid I, which is also known as (S)-1-[(E)-(2R,3S)-3-hydroxy-2,4-dimethyl-dodec-4-enoyl]-pyrrolidine-2-carboxylic acid, remain unclear. However, the parent nucleus in its chemical structure is pyrrolidine-2-carboxylic acid, which acts as an ionotropic glutamate receptor for some metabolites (Kayser et al., 2020). In the early phase of puberty, the hypothalamus, pituitary, and gonad axis are re-activated (Abreu and Kaiser, 2016), which is controlled by several neuroendocrine and metabolic factors (Walvoord, 2010; Plant, 2015). In addition, the GnRH pulse source is under both excitatory and inhibitory control, so, at the onset of puberty, the excitatory signal increases while the inhibitory signal decreases (Uenoyama et al., 2014). One neurotransmitter that stimulates GnRH pulse generation is glutamate (Farello et al., 2019). The pathogenesis of CPP has recently been linked to three genes: KISS1 (Silveira et al., 2010) encoding kisspeptin, its KISS1R receptor (Teles et al., 2008), and MKRN3, a gene deemed to act as a hypothalamic repressor on the gonadal axis (Farello et al., 2019). However, the exact mechanism of metabolites and genes remains to be studied and verified. A parent nucleus metabolite can decrease NH₃ production by inhibiting the deamination of amino acids (Floret et al., 1999). In addition, some complexes not only exhibit marginal antimicrobial activity but also have good cytotoxic activity (Aiyelabola et al., 2017). Therefore, the role of tumonoic acid I in CPP warrants further investigation.

Palmitic amide is a primary fatty acid amide derived from palmitic acid (C16:0) (Whipple et al., 2021). Its plasma levels have been reported to increase with increasing age among patients who are overweight 2 (Guo et al., 2021). Additionally, palmitic amide promotes infection by enhancing host glycolysis (Zhang et al., 2021). As shown in Figure 3A, PE [19:1(9Z)0:0] was positively correlated with tumonoic acid I, PC (16:0/0:0), and Phe-Phe but was negatively correlated with trans-trans-muconic acid.

Linoleic acid–biotin is the combination of linoleic acid and biotin. As a precursor of arachidonic acid (Castrejón-Téllez et al., 2019), linoleic acid is an essential omega-6 (or n-6) polyunsaturated fatty acid (PUFA) (Vannice and Rasmussen, 2014; Jandacek, 2017). A recent review indicated that inflammatory effects can be decreased by reducing the intake of linoleic acid (Naughton et al., 2016). In addition, linoleic acid can promote tumor growth in some species and under some conditions (Jandacek, 2017).

Water-soluble biotin is required for normal cellular functions, growth, and development (Said, 2009). Biotin plays diverse roles in normal immune function (áez-Saldaña et al., 1998) and cell proliferation (Manthey et al., 2002). Therefore, its deficiency can lead to various clinical abnormalities, including growth retardation, neurological disorders, and dermatological abnormalities (León-Del-Río, 2019).

This study revealed 14 pathways (the top two pathways are aminoacyl–tRNA biosynthesis and phenylalanine, tyrosine, and tryptophan biosynthesis) significantly associated with CPP. The enrichment scores of the aminoacyl–tRNA biosynthesis pathway were the highest, suggesting its involvement in the occurrence and/or progression of CPP. In a study on the complexity of metabolic changes and nutrient remobilization during leaf senescence of tobacco, dynamic changes in the biosynthesis of aromatic amino acids, such as phenylalanine, tyrosine, and tryptophan, have been observed (Li et al., 2017). Therefore, an in-depth investigation should be performed to identify metabolic pathways underlying the pathogenesis of CPP and associated with its treatment.

This study has some limitations. The age of female children in all groups was similar, and the associated diseases (except for injuries), dietary supplements, and diet differences were not examined. Furthermore, the number of samples used for screening key metabolites was relatively small (healthy female children, n = 30; patients with CPP, n = 31; patients treated with two drugs, n = 29) but met the requirements for human metabolomic analysis. However, the results should be validated in independent, large sample studies in the future. Moreover, female children with CPP can be primarily diagnosed based on the early appearance of secondary sexual characteristics, advanced bone age, gonadal enlargement and pubertal levels of serum gonadotropin, and sex hormones. This study demonstrated complementary methods that may improve clinical prediction and help understand the progression of CPP in female children.