Abstract

Background

   Precocious puberty (PP) in girls is traditionally defined as the onset
   of breast development before the age of 8 years. The specific
   biomarkers of premature thelarche (PT) and central precocious puberty
   (CPP) girls are uncertain, and little is known about their metabolic
   characteristics driven by perfluorinated compounds (PFCs) and clinical
   phenotype. This study aimed to screen specific biomarkers of PT and CPP
   and elucidate their underlying pathogenesis. The relationships of
   clinical phenotype-serum PFCs-metabolic characteristics were also
   explored to reveal the relationship between PFCs and the occurrence and
   development of PT and CPP.

Methods

   Nuclear magnetic resonance (NMR)-based cross-metabolomics strategy was
   performed on serum from 146 PP (including 30 CPP, 40 PT, and 76
   unspecified PP) girls and 64 healthy girls (including 36 prepubertal
   and 28 adolescent). Specific biomarkers were screened by the uni- and
   multivariate statistical analyses. The relationships between serum PFCs
   and clinical phenotype were performed by correlation analysis and
   weighted gene co-expression network analysis to explore the link of
   clinical phenotype-PFCs-metabolic characteristics in PT and CPP.

Results

   The disordered trend of pyruvate and butyrate metabolisms (metabolites
   mapped as formate, ethanol, and 3-hydroxybutyrate) were shared and kept
   almost consistent in PT and CPP. Eight and eleven specific biomarkers
   were screened for PT and CPP, respectively. The area under curve of
   specific biomarker combination was 0.721 in CPP vs. prepubertal, 0.972
   in PT vs. prepubertal, 0.646 in CPP vs. prepubertal integrated
   adolescent, and 0.822 in PT vs. prepubertal integrated adolescent,
   respectively. Perfluoro-n-heptanoic acid and perfluoro-n-hexanoic acid
   were statistically different between PT and CPP. Estradiol and
   prolactin were significantly correlated with PFCs in CPP and PT.
   Clinical phenotypes and PFCs drive the metabolic characteristics and
   cause metabolic disturbances in CPP and PT.

Conclusions

   The elevation of formate, ethanol, and 3-hydroxybutyrate may serve as
   the early diagnostic indicator for PP in girls. But the stratification
   of PP still needs to be further determined based on the specific
   biomarkers. Specific biomarkers of CPP and PT exhibited good
   sensitivity and can facilitate the classification diagnosis of CPP and
   PT. PFC exposure is associated with endocrine homeostasis imbalance.
   PFC exposure and/or endocrine disturbance directly or indirectly drive
   metabolic changes and form overall metabolic network perturbations in
   CPP and PT.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12916-023-03032-0.

   Keywords: Premature thelarche, Central precocious puberty, Specific
   biomarkers, Metabolic network, Perfluorinated compounds

Background

   Pubertal timing is usually regulated by complex interplay of genetic,
   environmental, nutritional, and epigenetic factors. Therefore, the
   criteria for normal pubertal timing and thus the definition of
   precocious puberty are hard to determine. Precocious puberty (PP) in
   girls is traditionally defined as the onset of breast development
   before the age of 8 years [[41]1]. Its underlying pathophysiology may
   be gonadotropin-releasing hormone (GnRH)-dependent for central
   precocious puberty (CPP) girls or GnRH-independent for premature
   thelarche (PT) girls. CPP is mainly induced by the continuous pulse
   secretion of GnRH to prematurely activate the
   hypothalamic-pituitary-gonadal (HPG) axis; however, the exact
   mechanisms remain unclear. The main clinical manifestation of the PT
   girls is simple breast development due to exposure to the peripheral
   estrogen environment. When PT is accompanied by the significant advance
   growth of bone age, it is more likely to evolve into secondary CPP. CPP
   can lead to short- and long-term complications in girls, including
   increased risk of psychosocial distress, short stature, obesity,
   cardiovascular disease, and type 2 diabetes in adulthood [[42]2].
   Therefore, it is vital to understand the etiology of PT and CPP for
   accurate diagnosis and prompt intervention.

   Some researchers tried to quantify PP with the help of clinical
   phenotype such as luteinizing hormone (LH), follicular-stimulating
   hormone (FSH), and estradiol to determine the index threshold for CPP
   diagnosis, but it is still confronted with great controversy and
   challenge at this moment [[43]3, [44]4]. Some evidences have indicated
   the changes of metabolic profile during puberty. Qi et al. found that
   catecholamine metabolic pathway, tryptophan metabolic pathway, and TCA
   cycle were disturbed in CPP girls by GC/LC-MS-based urinary
   metabolomics analysis [[45]5]. Yang et al. used LC-MS technology to
   characterize the urinary metabolomes of CPP girls and found that amino
   acids, especially aromatic amino acids, were closely related to the
   pathogenesis of CPP by activating the HPG axis and inhibiting the
   hypothalamic-pituitary-adrenal axis [[46]6]. However, the clinical
   differential diagnosis of CPP and PT is still in a vague interface, and
   the lack of powerful molecular biomarkers is a long-term bottleneck in
   the clinical diagnosis and evaluation of PP.

   Recently, ubiquitous exposures to polyfluorinated compounds (PFCs) have
   attracted concerns regarding their possible harmful effects during
   critical periods of development in early-life and long-term
   consequences on health in consideration of their persistence and
   bioaccumulation potential. Massive researches have shown that PFCs can
   interfere with estrogen homeostasis and pose a risk of
   endocrine-disrupting effects [[47]7–[48]9], and they are association
   with dyslipidemia, renal function, and age at menarche [[49]7,
   [50]10–[51]12], but there is still inconsistency in the research
   results [[52]13] as well as certain gender differences [[53]14–[54]17].
   In addition, evidences have shown that PFCs can affect the HPG axis
   [[55]18, [56]19] or directly affect the gonad axis through their weak
   estrogen or antiandrogen effects to disrupt the development of puberty
   [[57]20]. However, the correlation research of PFC exposure with the
   occurrence and development of PP is still in its infancy [[58]19,
   [59]21, [60]22], and therefore extensive in-depth research and
   exploration is urgently required to clarify the exact response
   mechanism. The correlation analysis between PFCs and the clinical
   phenotype in girls with PP as well as the endogenous metabolites driven
   by PFCs will help to reveal the impact of PFCs on the occurrence and
   development of precocious puberty in girls and the preliminary
   mechanisms.

   Based on this, the serum metabolic profiles of prepubertal, PP, PT,
   CPP, and adolescent girls were characterized by one-dimensional nuclear
   magnetic resonance hydrogen spectrum (^1H-NMR) technology, and the
   metabolic differences and connections were analyzed by
   cross-metabolomics analyses, aiming to screen the specific biomarkers
   of CPP and PT. Furthermore, to reveal the effect of PFCs on the
   occurrence of CPP and PT, the metabolic modules driven by PFCs and
   clinical phenotype were identified by weighted gene co-expression
   network analysis (WGCNA).

Methods

Subject selection and sampling

   The children were enrolled from the Department of Child Health, Women
   and Children’s Hospital, School of Medicine, Xiamen University. A total
   of 146 PP (including 30 CPP, 40 PT, and 76 unspecified PP) girls were
   enrolled at their first visit. The inclusion criteria of the patients
   are shown in detail in Fig. [61]1 according to the clinical guidelines
   and the related literatures [[62]23–[63]25]. In addition, 64 healthy
   girls were recruited as a control group for metabolic comparison, who
   were divided into 36 prepubertal and 28 adolescent girls based on their
   developmental status. Relevant clinical phenotypes were collected
   during the clinical examination. Morning fasting serum sample was
   collected from each girl through a clinical standard procedure and
   stood for 30 min, then centrifuged at 1000g for 10 min at 4 °C. The
   serum supernatant was transferred to a new centrifuge tube and stored
   at −80 °C until analysis.

Fig. 1.

   [64]Fig. 1
   [65]Open in a new tab

   Inclusion and screening process for study cohorts. *: Depending on the
   method of measurement. The detection method of basal LH and GnRH
   stimulation test in this study was immunochemiluminometric. LH:
   luteinizing hormone; FSH: follicular-stimulating hormone; BA: bone age
   (at the time of diagnosis); CA: chronologic age (at the time of
   diagnosis)

Sample preparation, ^1H-NMR spectra acquisition and processing

   All serum samples were thawed at 4 °C, and 400 μL of serum was mixed
   with 200 μL of 60 mM phosphate buffer (pH 7.4, in 0.9% deuterated
   saline solution) and then vortexed for 10 s. After being centrifuged at
   13,000g for 10 min at 4 °C, 550 μL of supernatant was transferred into
   a 5-mm NMR tube for ^1H-NMR spectral acquisition.

   The ^1H-NMR spectra of serum samples were obtained on a 600-MHz Bruker
   Advance nuclear magnetic resonance (NMR) spectrometer (Bruker BioSpin,
   Germany) equipped with a triple resonance cryogenic probe operating at
   600.13 MHz and 298.0 K. A typical water-suppressed
   Carr-Purcell-Meiboom-Gill (CPMG, [RD-90°-(τ-180°-τ)[n]-ACQ]) pulse
   sequence with a spectral width of 12,019.2 Hz, an acquisition time of
   1.36 s, a relaxation delay of 4.0 s, a scan accumulation of 64 times,
   and a data point of 16 K was adopted to acquire ^1H-NMR spectra.

   Spectral processing was performed on MestReNova (version 14.1.1,
   Mestrelab Research S.L., Spain). All the free induction decays were
   zero-filled to 64 K data points and multiplied by an exponential
   function of 1.0 Hz line-broadening factor. The ^1H-NMR spectra were
   manually phased, and baseline corrected, and then referenced to the
   doublet of endogenous lactate at δ1.33 after Fourier transformation.
   The spectral regions of δ4.70–δ5.17 and δ5.50–δ6.00 were removed to
   eliminate the interference of residual aquatic and urea signals. The
   remainder spectral regions (δ0.55–δ8.60) were integrally segmented into
   discrete regions of 0.002 ppm. To reduce the concentration difference
   between the samples, the obtained NMR spectral data were normalized to
   the total integrated area.

LC-MS/MS detection of serum PFCs

   The serum PFCs of 40 PT and 30 CPP girls were measured by LC-MS/MS
   technique, and eleven kinds of PFCs, including perfluoro-n-octanoic
   acid (PFOA), potassium perfluoro-1-octanesulfonate (PFOS),
   perfluoro-n-butanoic acid (PFBA), perfluoro-n-undecanoic acid (PFUnDA),
   perfluoro-n-dodecanoic acid (PFDoDA), potassium
   perfluoro-1-butanesulfonate (PFBS), perfluoro-n-decanoic acid (PFDA),
   perfluoro-n-heptanoic acid (PFHpA), perfluoro-n-hexanoic acid (PFHA),
   potassium perfluoro-1-hexanesulfonate (TFHSA), and perfluoro-n-nonanoic
   acid (PFNA), were detected according to the references [[66]26,