Circulating matrix GLA protein during puberty in healthy and central precocious puberty children

Abstract

Objective

Matrix GLA protein (MGP) is involved in the regulation of extracellular matrix mineralisation, including bone formation. The peripubertal years are critical for linear growth and acquisition of bone mass. Emerging evidence suggests that plasma concentrations of dephosphorylated undercarboxylated MGP (dp-ucMGP) may vary across pubertal stages, potentially reflecting its role in bone remodelling.

Methods

This study examines longitudinal changes in circulating plasma dp-ucMGP in two cohorts: 111 healthy Danish children (52% girls) aged 6–15 years from the Copenhagen Puberty Study and 15 female patients with central precocious puberty (CPP) before, during, and after GnRH analogue treatment.

Results

Age- and sex-specific reference curves were generated for healthy children and adolescents. Dp-ucMGP increased significantly with age and pubertal progression in both sexes, while it declined from mid-puberty exclusively in boys. In girls only, dp-ucMGP SD scores were associated with body weight and BMI. In girls with CPP, dp-ucMGP concentrations were lower compared with healthy puberty-matched girls (Tanner B2: P = 0.016, Tanner B3: P = 0.0072). Initiation of GnRH agonist in CPP resulted in a significant increase in dp-ucMGP (−1.21 SD (baseline) vs 0.30 SD (3 months), P = 0.0048) but plateaued after 12 months of treatment and further declined towards baseline at treatment cessation.

Conclusion

We show that circulating dp-ucMGP concentrations increase with age and pubertal progression in healthy children, are lower in CPP patients, and are suppressed by GnRH agonist. Altogether, our findings support that circulating concentrations of dp-ucMGP reflect its role in bone remodelling and linear growth during puberty.

Plain language summary

In this study, we show that matrix GLA protein (MGP), a regulator of bone formation, increases with pubertal progression in healthy children but is suppressed when puberty is pharmacologically blocked in girls with early puberty. Our findings provide new insights into the potential role of MGP in bone growth during puberty.

Introduction

Puberty involves a coordinated neuroendocrine process that marks the transition from childhood to adulthood. It is initiated by reactivation of the hypothalamic–pituitary–gonadal (HPG) axis, leading to increased pulsatile secretion of gonadotrophin-releasing hormone (GnRH), followed by the release of luteinising hormone (LH) and follicle-stimulating hormone (FSH), resulting in increasing levels of sex steroids. These hormonal changes drive the development of secondary sexual characteristics, reproductive maturation, and marked acceleration in linear growth, with an increase in testicular volume in boys and breast development in girls being the first signs of pubertal onset (1).

In children with central precocious puberty (CPP), the HPG axis is reactivated prematurely, leading to early pubertal onset and accelerated linear growth for age. The early rise in sex steroids stimulates the pubertal growth spurt and ultimately premature fusion of the epiphyseal plates, which may consequently compromise attainment of the expected final height. One rationale for initiating treatment with GnRH agonists, blocking the activated HPG axis, in children with CPP, is to improve their ability to achieve their genetic height potential (2).

Longitudinal growth of long bones takes place in the epiphyseal growth plate, where chondrocytes proliferate and differentiate into hypertrophic chondrocytes, coordinating vascular invasion, cartilage mineralisation, and its replacement by osteoblast-derived matrix. Parallelly, bone remodelling continues through a balanced process of bone formation and resorption (3). Sex steroids and growth hormone (GH), acting through insulin-like growth factor I (IGF-I), drive this process through childhood and puberty (4, 5). Eventually, local oestradiol (E2) will contribute to the decline in chondrocyte proliferation rate, leading to growth plate fusion and arrest of longitudinal growth (3, 6).

The vitamin K-dependent protein osteocalcin (or bone GLA protein) is known to promote the build-up of bone matrix by stimulating collagen synthesis (7) and has been identified as a sensitive marker of linear bone growth in healthy children (8). Another, less well-characterised vitamin K-dependent protein, relevant to bone matrix formation, is matrix GLA protein (MGP) (9). MGP is expressed by chondrocytes, fibroblasts, and vascular smooth muscle cells, where it exerts diverse regulatory functions primarily due to its calcium-binding capacity (10). MGP expression by adipocytes and involvement in fat metabolism has also been suggested (11, 12). MGP promotes bone formation by stimulating osteoblast proliferation (13), but it also exerts inhibitory effects on extracellular matrix mineralisation (14, 15), and MGP deficiency is known to cause inappropriate calcification of extracellular matrix involving multiple tissues (16, 17). The biological activity of MGP is governed by two post-translational modifications: carboxylation and serine phosphorylation, yielding four different molecular forms. Fully modified p-cMGP is predominantly tissue-bound and represents the fully activated form, participating in the regulation of extracellular matrix mineralisation; whereas dephosphorylated–undercarboxylated MGP (dp-ucMGP) is considered biologically inactive, but is measurable in plasma (10, 14, 18, 19). In a cross-sectional study investigating effects on bone mineral density in children with chronic kidney disease, treated with anticoagulation therapy, levels of dp-ucMGP were found to increase with age and stages of pubertal development (20). However, the potential role of MGP in bone modelling during normal pubertal growth remains incompletely characterised, and available data on circulating dp-ucMGP levels in healthy paediatric populations are limited (21). Therefore, this study aims to investigate circulating dp-ucMGP longitudinally during pubertal development in healthy boys and girls enrolled in the Copenhagen Puberty Study, as well as in young girls diagnosed with central precocious puberty (CPP) before, during, and after treatment with a GnRH analogue.

Materials and methods

Subjects

This study was part of the Copenhagen Puberty Study (22, 23), a combined cross-sectional and longitudinal population-based cohort study in healthy Danish children and adolescents.

Dp-ucMGP plasma concentrations were assessed in 111 children (52% girls) from the longitudinal part of the primary study. Participants were clinically examined and had blood samples drawn every 6 months between 2007 and 2014, both before and after pubertal onset (aged 6–15 years), as previously described (24). The included children were generally healthy and had no major comorbidities. Three children were receiving daily treatment with inhaled corticosteroids.

The second study group consisted of 15 girls with CPP from a prospective clinical study (25). Premenarchal patients presenting with breast development before age 9 years were recruited from the outpatient clinic at the Department of Growth and Reproduction, Copenhagen University Hospital, from May 2008 to September 2009. CPP was defined, in accordance with guidelines (26), by breast budding before the age of 8 years, peak LH > 5 IU/L and/or peak LH-to-peak FSH ratio > 0.66, following GnRH-agonist stimulation, and nonpathological brain MRI. Two girls presented with pubertal onset at age 8.3 years and were offered treatment equivalent to that provided to girls diagnosed with CPP due to rapid pubertal progression. All patients will be referred to as CPP in this paper. Median age of pubertal onset in girls with CPP was 7.7 years (range: 7.2–8.6 years). Patients were treated with a long-acting GnRH agonist (Procren Depot, Abbott Scandinavia AB; leuprolide acetate 3.75 mg), with subcutaneous injections every 28th day. Median age at initiation of treatment was 8.9 years (range: 7.5–10.0 years).

Blood samples were collected before treatment initiation and at re-examinations after 3 and 12 months of treatment, and 6 months post-treatment, as previously described (27).

Clinical examination

Pubertal stages were evaluated by trained physicians according to the classifications of Marshall and Tanner (28, 29) with evaluation of breast development (B1–B5) and testicular volume (Tvol) (mL). Pubertal onset was defined as persistent stage B2 (or higher) in girls and Tvol>= 4 mL in boys. Age of pubertal onset was calculated as the mean between the age of first examination with Tanner stage B2+ or Tvol>= 4 mL and the age of examination before this. Years from pubertal onset were calculated as the difference between age at examination and age of pubertal onset in years (y). Anthropometric data, weight, height, and body mass index (BMI; (weight (kg)/(height (m))²)) were measured and further expressed as sex- and age-adjusted standard deviations (SD).

Laboratory analysis

Blood samples were drawn from an antecubital vein between 08:00 and 10:00 h for the healthy population and in outpatient clinic opening hours between 08:00 h and 14:00 h for girls with CPP. Serum and plasma samples were stored at −20°C.

Dp-ucMGP levels were measured in potassium–EDTA–plasma samples using a non-commercial immunoassay of IDS-iSYS Multi-Discipline Automated System (IDS-iSYS InaKtif MGP) in 2016 according to the manufacturer´s protocol (IDS Immunodiagnostic Systems Deutschland GmbH|Herriotstr. 1, 60528, Germany). The non-commercial assay is now automated and commercially available (IS-4700/IS-4730) with a reported limit of quantification (LOQ) at 300 pmol/L and limit of detection (LOD) at 200 pmol/L. Neither the non-commercial 2016 kit nor the following identical commercially available kit has been validated in a paediatric cohort, according to the manufacturer. Plasma dp-ucMGP was measured in 445 samples in total: 236 male samples (from 53 unique boys), and 209 female samples (from 58 unique girls). The median number of dp-ucMGP measurements per individual in the healthy population was 4.0 (IQR: 3.0; 5.0) for boys, ranging from 1 to 7 in samples, and 5.0 (IQR: 4.0; 5.0) for girls, ranging from 1 to 5 in samples. The sample range was 15–786 pmol/L. In total, 202 (85.6%) blood samples in healthy boys and 145 (69.4%) in healthy girls were below LOQ, but all yielded a read-out. About 232 (52.1%) samples had dp-ucMGP below LOD. Fifty-nine (95.2%) of dp-ucMGP measurements in girls diagnosed with CPP were below LOQ, and 44 (71%) were below LOD. All measurements of dp-ucMGP yielding a read-out were included in final analyses.

Serum levels of E2 were determined by RIA (Pantex) with an LOD of 18 pmol/L, and intra- and interassay coefficient of variation (CV) values were 7.5 and 12.3%, respectively. Serum testosterone levels were measured with DPC Coat-A-Count RIA kit (Diagnostic Products Corp) with LOD of 0.23 nmol/L, and the intra- and interassay CV values were 7.6 and 8.6%, respectively, as previously described (24). IGF-I was analysed using immunoassay IDS-iSYS IGF-I with an LOD of 10 ng/mL, and interassay CV was below 7.2%, as previously described (30). All analyses were performed in-house at the Department of Growth and Reproduction, Rigshospitalet, Denmark.

Statistical analysis

Longitudinal data from 111 healthy children were analysed, with each child contributing one or more data points depending on the number of visits. Due to multiple measurements for each individual, differences in plasma concentrations of dp-ucMGP across sex, chronological age, and pubertal stages in the healthy population were analysed using a left-censored linear mixed-effects model (with the lme4cens package in R), to account for measurements below the reported LOD, with a random intercept for subject to account for repeated measurements. Measurements below the LOD were included in the analyses as being below the respective LOD threshold, rather than using their numerical assay read-out, excluding them, or assigning them the exact LOD value. Analyses of sex differences in dp-ucMGP concentrations were adjusted for age. Bonferroni–Holm correction was applied in group comparisons of pubertal stages to account for multiple comparisons.

Sex- and age-specific reference charts for dp-ucMGP were developed, and standard deviations (SD) for dp-ucMGP, IGF-I, and anthropometrics were calculated using the generalised additive models for location, scale and Shape (GAMLSS). Data were summarised in three smoothed age-dependent curves: L, M, and S, where L adjusts for age-dependent skewness, M corresponds to the age-dependent median, and S approximates the age-dependent coefficient of variation. This method is based on the Box–Cox power transformation, which transforms the data to follow a Gaussian distribution for each age, as previously described (30, 31).

Associations between crude values and SD dp-ucMGP and anthropometric measurements, sex steroids, and IGF1 were also examined using a left-censored linear mixed-effects model. Censoring for SD values was performed at the corresponding age- and sex-specific SD value of the LOD.

Intraclass correlation coefficients (ICCs) were estimated from left-censored random-intercept linear mixed-effects models fitted by maximum likelihood, with measurements below the LOD treated as censored observations.

Group comparisons between baseline measures from girls diagnosed with CPP and healthy-matched girls were performed using generalised estimated equations (GEEs). Due to multiple measurements for each individual, reported means for statistical comparisons are estimated as marginal means (EMMs) for anthropometrics and E2, whereas dp-ucMGP was reported as median with interquartile range (IQR) due to non-Gaussian distribution. In addition, the proportions of measurements above versus below the LOD for both dp-ucMGP concentrations and dp-ucMGP SD were compared between girls with CPP and healthy girls using Fisher’s exact test. Differences in the proportion of dp-ucMGP values (crude values and SD) above LOD in girls diagnosed with CPP before, during, and after treatment were likewise assessed using Fisher’s exact test.

All statistical analyses were performed using R Statistical Software (version 4.5.0; R Core Team, 2025). P-values < 0.05 were considered statistically significant.

Ethical considerations

The Copenhagen Puberty Study was approved by the local ethical committee (nos KF01282214 and KF112006-2033). The primary study is registered in ClinicalTrials.gov (NCT01411527), and data are stored according to European GDPR (P-2020-1063). All participants and their parents gave informed consent before they participated in the study.

Results

Plasma levels of dp-ucMGP in a cohort of healthy boys and girls

Baseline characteristics at first examinations of the healthy population including dp-ucMGP concentrations are presented in Table 1, and the pubertal stages are summarised in Table 2. Ages of pubertal onset (median (IQR)) were 11.6 (10.9; 12.01) in boys and 10.5 (9.7; 11.4) in girls.

Table 1.

Baseline characteristics of healthy population stratified by sex. Data are presented as median (IQR) or n (%).

	Boys	Girls
Total n (%)	53 (47.7%)	58 (52.3%)
Dp-ucMGP (pmol/L)	147 (94; 221)	187 (143; 227)
Anthropometrics
Age (years)	10.8 (10.2; 10.7)	10.3 (9.3; 11.1)
Height (cm)	146.1 (142.5; 151.4)	142.5 (136.6; 145.9)
Height (SD)	−0.23 (−0.75; 0.53)	−0.39 (−0.86; 0.32)
Weight (kg)	37.5 (34.3; 40.3)	33.2 (30.4; 36.8)
Weight (SD)	−0.01 (−0.35; 0.55)	−0.16 (−0.82; 0.25)
BMI	17.2 (16.1; 18.6)	16.5 (15.3; 18.2)
BMI (SD)	0.03 (−0.51; 0.72)	−0.10 (−0.77; 0.50)

Table 2.

Pubertal stage at baseline.

	Boys		Girls
	TV, mL	n (%)	BS	n (%)
Prepubertal	1–3	41 (77.4%)	B1	35 (60.3%)
Pubertal	4–5	12 (22.6%)	B2	12 (20.7%)
	6–10	0 (0%)	B3	9 (15.5%)
	11–15	0 (0%)	B4	1 (1.7%)
	>15	0 (0%)	B5	1 (1.7%)

TV, testicular volume; BS, breast stage.

Dp-ucMGP levels increased significantly with chronological age, through childhood in both boys and girls (boys: β = 43.7, 95% CI: (25.5; 61.9); P < 0.0001; girls: β =54.5, 95% CI: (40.1; 69.0); P < 0.0001; Fig. 1A and B). Longitudinal dp-ucMGP concentrations (median (IQR)) were lower in healthy boys compared with girls (β = −85.5, 95% CI: (−124.7; −46.2); P < 0.0001), when adjusting for age.

Figure 1.

Dp-ucMGP in healthy children increases with age and pubertal progression. Each line represents the longitudinal trajectory of one child, dots represent individual measurements. The solid line represents a locally estimated scatterplot smoothing (LOESS) curve fitted to the data. (A) Levels of dp-ucMGP (pmol/L) by age (years) for boys stratified by testicular volume (Tvol) (mL). (B) Levels of dp-ucMGP (pmol/L) by age (years) for healthy girls stratified by Tanner stages of breast development. (C) Levels of dp-ucMGP (pmol/L) by years from age at pubertal onset (years) in boys stratified by Tvol (mL). (D) Levels of dp-ucMGP (pmol/L) by years from age at pubertal onset (years) in girls stratified by Tanner stages of breast development.

Dp-ucMGP concentrations increase with pubertal progression in a healthy paediatric population

When assessing dp-ucMGP longitudinally in relation to pubertal onset, the most pronounced increase was observed around the time of pubertal onset; however, no sharp peak at pubertal onset was evident (Fig. 1C and D).

Dp-ucMGP levels increased significantly from prepubertal stage (Tvol 1–3 mL) through the following stages after pubertal onset: (4–5 mL: β = 70.5, 95% CI: (29.3; 111.8); P = 0.0024) and (6–10 mL: β = 132.98, 95% CI: (88.04; 177.9); P < 0.0001) (Fig. 2A).

Figure 2.

Dp-ucMGP across pubertal stages in healthy children. Box plots show median and IQR. Individual measurements are shown as dots. Statistical analysis: left-censored linear mixed-effects model (lmercens), Bonferroni–Holm-adjusted. (A) Levels of dp-ucMGP (pmol/L) by Tvol (mL) in boys. (B) Levels of dp-ucMGP (pmol/L) by Tanner breast stages (B1–B5) in girls.

Notably, from mid-puberty in males, dp-ucMGP levels were not statistically different from prepubertal levels (Fig. 2A). In girls, there is a significant increase in dp-ucMGP concentrations between prepubertal girls (Tanner stage B1) and all the following pubertal stages (B2: β = 74.0, 95% CI: (37.03; 110.9), P < 0.001; B3: β = 111.02, 95% CI (72.5; 149.6), P < 0.0001; B4: β = 180.2, 95% CI: (121.3; 239.1); P < 0.0001; B5: β = 270.1, 95% CI: (132.1; 408.2); P < 0.001) (Fig. 2B).

Dynamics of dp-ucMGP across age and sex, and tracking through childhood

Healthy girls and boys had comparable progression of dp-ucMGP concentrations through childhood with a widening of the normal range in adolescence, reflecting a considerable inter-individual variation with increasing age (Fig. 3A and B). Notably, in adolescent males, the plasma concentrations seemingly plateaued, which was not mirrored in the females. Intraclass correlation coefficients (ICC) of dp-ucMGP SD for each child showed moderate within-subject consistency throughout the course of the study. The ICC of dp-ucMGP SD was 0.44 (95% CI: 0.25; 0.62) for boys and 0.46 (95% CI: 0.22; 0.69) for girls.

Figure 3.

(A) Dp-ucMGP (pmol/L) concentrations by age (years) for boys (green) and (B) for healthy girls (pink) and girls with CPP (blue), including reference lines for +2 SD, +1 SD, median, −1 SD, and −2 SD, respectively. Reference material is only based on data from the healthy population.

In girls, dp-ucMGP SD increases with increasing weight and BMI, but not IGF-I

In girls, dp-ucMGP SD was positively associated with weight SD (β = 0.26, 95% CI (0.04; 0.48), P = 0.021, Fig. 4A) and BMI SD (β = 0.25, 95% CI: (0.056; 0.44); P = 0.011, Fig. 4B), whereas no association was seen in boys (data not shown). No significant association was found between dp-ucMGP and IGF-I, regardless of sex, assessed as both crude values and SD. Furthermore, no significant associations were found with height SD, E2, or testosterone, regardless of sex (data not shown).

Figure 4.

Left-censored linear mixed-effects model (lmercens) associations between dp-ucMGP SD and weight SD (A) and BMI SD (B) in healthy girls. Each cluster of the same coloured dots represents measurements from the same child.

Plasma levels of dp-ucMGP are significantly lower in girls with CPP before treatment, compared with healthy girls matched on Tanner stage

Anthropometrics and levels of E2 and dp-ucMGP in girls with CPP before treatment initiation, compared with healthy girls matched on Tanner stages of breast development, are summarised in Table 3. Of the 15 girls diagnosed with CPP, 11 had blood samples available at baseline.

Table 3.

Anthropometrics, E2, and dp-ucMGP for girls with CPP compared with healthy girls, matched by Tanner stages of breast development. Measurements are reported as estimated mean (95% CI) for anthropometrics and E2 and as median (IQR) for dp-ucMGP and dp-ucMGP-SDS. Anthropometric measurements and dp-ucMGP concentrations were compared between girls with CPP and healthy girls, matched on Tanner stages of breast development. For girls with CPP, only baseline values before treatment initiation are presented. For healthy girls, all measurements classified as Tanner breast stage B2 or B3 are presented. Also, 18 of the healthy girls with B2 and 18 with B3 contributed two or three measurements as they were followed longitudinally.

	Female CPP patients		Healthy girls		P-value*
	B2	B3	B2	B3	B2	B3
Samples (n)	5	6	63	59
Individuals (n)	5	6	40	38
Age (y)	8.0 (7.5; 8.5)	9.0 (8.7; 9.2)	10.8 (10.4; 11.1)	11.7 (11.4; 12)	<0.0001	<0.0001
Height (cm)	133 (127; 139)	139 (134; 144)	145 (143; 146)	151 (149; 153)	<0.001	<0.0001
Height (SD)	0.60 (−0.27; 1.48)	0.56 (−0.2; 1.3)	−0.25 (−0.55; 0.06)	−0.04 (−0.41; 0.33)	0.07	0.16
Weight (kg)	31.9 (27.3; 36.4)	33.6 (30.0; 37.1)	36.0 (34.3; 37.8)	40.7 (38.6; 42.9)	0.09	0.00051
Weight (SD)	0.91 (0.15; 1.68)	0.57 (0.04; 1.10)	−0.19 (−0.49; 0.11)	−0.03 (−0.35; 0.29)	0.009	0.056
BMI	17.8 (16.6; 18.9)	17.3 (16.3; 18.2)	17.2 (16.5; 17.8)	17.8 (17.1; 18.5)	0.37	0.38
BMI (SD)	0.90 (0.39; 1.40)	0.49 (0.07; 0.91)	−0.08 (−0.40; 0.24)	−0.007 (−0.34; 0.32)	0.002	0.07
E2 (pmol/L)	43 (3; 68)	72 (32; 162)	45 (38; 55)	74 (61; 91)	0.8	0.94
Dp-ucMGP (pmol/L)	130 (120–179)	76 (38–87)	219 (155–339)	248 (173–326)
Values > LOD	0/5	0/6	37/63	34/59	0.016	0.009
Dp-ucMGP (SD)	−0.25 (−0.63–0.36)	−1.66 (−2.49 to −1.40)	−0.04 (−0.64–0.94)	0.01 (−0.81–0.76)
Values > LOD	0/5	0/6	37/63	35/59	0.016	0.0072

^*P-values and 95% confidence interval generated by generalised estimating equations (GEEs) with Gaussian distribution for age, height (cm and SD), weight (kg and SD), BMI (SD) and gamma distribution and log-link for E2 concentrations. P-values for proportions of dp-ucMGP values (pmol/L and SD) > LOD were generated using Fisher’s exact test. Significant P-values in bold.

Girls with CPP exhibited higher body weight SD than the healthy matched girls, though only significant for stage B2. Girls with CPP also had a higher BMI SD compared to healthy girls at stage B2 (Table 3).

In girls with CPP, baseline measurements of dp-ucMGP appeared to be in the lower range compared with healthy girls of similar age (Figs 3B and 5B). Furthermore, dp-ucMGP concentrations and SD scores were significantly lower in girls with CPP before treatment initiation compared with healthy girls matched on Tanner stage (Table 3). E2 concentrations did not differ significantly between girls with CPP and Tanner-matched healthy girls (Table 3).

Figure 5.

Levels of dp-ucMGP in girls with CPP before (baseline), during (3 and 12 months) treatment with GnRH analogue and 6 months after ending treatment (post-treatment). (A) Progression of dp-ucMGP in pmol/L. (B) Progression of dp-ucMGP SD. (C) Progression of E2 (pmol/L). Points connected by lines represent longitudinal measurements for each child. Dashed lines indicate that a measurement is missing between the connected points. Differences in the proportion of dp-ucMGP measurements above the limit of detection (LOD) across timepoints were compared with baseline (baseline vs 3 months, 12 months, and post-treatment) using Fisher’s exact test.

GnRH analogue suppresses MGP in girls with CPP

Dp-ucMGP continued to increase significantly in the girls with CPP, after three months of GnRH analogue treatment, compared with their baseline level (median baseline 115 pmol/L (IQR 76; 125) vs 3 months 260 pmol/L (IQR 160; 274)), corresponding to a change in SD (median baseline −0.91 SD (IQR −1.66; −0.44) vs 3 months 0.67 SD (IQR −0.18; 1.62)). When assessed as values > LOD vs < LOD, we confirmed this significant increase (measurements > LOD: baseline 0/11 vs 3 months 5/8, P = 0.0048) and SD scores > LOD (baseline 0/11 vs 3 months 5/8, P = 0.0048, Fig. 5A and B). Of note, once dp-ucMGP concentrations plateaued, levels stabilised from here on (3 vs 12 months; and 3 months vs post treatment both P = NS, Fig. 5A and B). Furthermore, 6 months after treatment cessation, levels of dp-ucMGP were still comparable to baseline levels (baseline vs post-treatment: P = NS).

Discussion

In our study, we report for the first time longitudinal measurements of dp-ucMGP levels increasing significantly with age and pubertal progression for both sexes in healthy children. During childhood, dp-ucMGP progressed comparably in boys and girls, but in adolescent males, plasma concentrations appeared to plateau, a pattern not seen in females, where levels continued to rise. In accordance with this, previous work has shown that other bone metabolic markers, such as BAP, osteocalcin, PINP, and ICTP, increased with pubertal progression towards a maximum at Tanner stage G4 in boys and B3 in girls (32); and we recently demonstrated that procollagen type I and type II biomarkers likewise increase throughout early pubertal progression and attain maximal levels around the time of peak height velocity (33). We show that in boys, dp-ucMGP levels increased significantly from prepubertal to more advanced stages, peaking around Tvol of 6–10, after which levels seemed to decline towards prepubertal concentrations. As reaching both final height and peak bone mass is achieved a few years later in boys (7, 34), one could expect levels of dp-ucMGP to remain higher in the boys. However, we found that dp-ucMGP levels in boys declined in late puberty. A similar pattern was reported with dp-ucMGP levels in the age range 20–39 years, resembling prepubertal levels (<10 years), though not stratified by sex (21). In contrast, we found that in healthy girls, dp-ucMGP levels showed an increasing trend throughout pubertal development, corresponding to findings by Thom et al. in children with chronic kidney disease (20). Another evident sex-specific association was dp-ucMGP with both body weight and BMI. This suggests that dp-ucMGP could be affected by load-dependent bone remodelling. Absolute body weight may act as a proxy for skeletal loading and muscle forces, which are key determinants of osteocyte-mediated bone turnover. However, MGP expression by adipocytes has been implied (11, 12), suggesting that its association with body weight may partly reflect fat mass. The female-specific association and the sex specific temporal changes in late puberty may therefore reflect differences in fat percentage, and further in sex steroid milieu, as E2 has been shown to modulate osteocyte mechanosensitivity and bone turnover (35), although this remains speculative. Rodent studies, supported by limited human data, have shown that growth-plate chondrocytes locally convert testosterone to either E2 or dihydrotestosterone in a sex-specific manner, underscoring potential sex differences at the cellular level that remain largely unexplored (36, 37).

Compared with healthy girls matched on Tanner stage, girls with CPP had significantly lower dp-ucMGP levels. Both anthropometrics and dp-ucMGP levels in the CPP group could theoretically be affected by the two girls with pubertal onset at age 8.3 years as they represent a transitional group, rather than strictly defined CPP; however, excluding these two individuals from the analyses did not alter significance (data not shown). Another hypothesis could be that early sustained exposure to E2 in girls with CPP lowers dp-ucMGP in plasma by shifting it to the tissue-bound activated form. However, this mechanism is unexplored.

Preclinical studies indicate that receptor activator of nuclear factor κB ligand (RANKL) reduces MGP mRNA expression in human aortic smooth muscle cells, an effect reversed by E2 administration; moreover, the expression decreased significantly after ovariectomy in rats but was restored with E2 (38). Supporting this relationship, low-dose E2 upregulated MGP in ovariectomised mice (13). In girls with CPP, dp-ucMGP concentrations continued to increase significantly from baseline through 3 months of GnRH treatment. Interestingly, after 12 months of treatment, plasma levels plateaued and remained stable, with levels comparable to baseline even 6 months after treatment cessation. This may indicate that dp-ucMGP concentrations decline towards baseline levels when pubertal progression is inhibited, and E2 levels are lowered. This ‘delayed’ effect of GnRH on dp-ucMGP may suggest that once the growth spurt is initiated, it is not rapidly slowed down. Supporting this, a longitudinal study in children with CPP treated with GnRH agonist reported that bone metabolic markers osteocalcin, PICP, and ICTP decreased after 6 months of treatment (39). However, dp-ucMGP concentration may not accurately reflect biologically active p-cMGP within the growth plates, and the delayed GnRH effects on dp-ucMGP might be discordant locally in epiphyseal cartilage. No significant increase in dp-ucMGP was found 6 months post-treatment, as opposed to E2 levels, questioning whether MGP is under direct E2 regulation. There is currently no direct evidence that either androgens or IGF-I regulate MGP expression or circulating dp-ucMGP levels, and in this study, no associations with serum IGF-I or testosterone were seen. MGP is so far mainly known to be regulated by the availability of vitamin K and local mineralisation stimuli such as bone morphogenetic protein-2 (BMP-2), phosphate, and calcium (40). Despite total serum calcium being stable across Tanner stages in both boys and girls, serum phosphorus has been shown to peak at Tanner stage 2 in girls and 3 in boys (41), which mirrors dp-ucMGP in our study. High levels of dp-ucMGP may reflect poor vitamin K status. As the included children were generally healthy, information on vitamin supplement use was not collected. Consequently, a potential influence of vitamin K status on dp-ucMGP levels cannot be excluded. Nutritional status should, therefore, be considered in future studies, particularly when analysing circulating dp-ucMGP concentrations in non-healthy children.

One limitation of our study is the numerous values <LOQ. However, others using assays comparable to that used in this study have reported similar dp-ucMGP levels in healthy children, and others found that 39% of dp-ucMGP samples in healthy adults were <LOQ, with all measured values included in the analyses (19, 21). All observations <LOD, both crude values and SD scores, were accounted for as left-censored values in the linear mixed-effects models, rather than treated as exact measured concentrations, or excluded altogether. The high degree of censoring reduces information on the distribution and variability of biomarker concentrations at the lower end of the measurement range; however, the censored approach avoids introducing artificial precision. Intraclass correlation coefficients of dp-ucMGP SD demonstrated moderate within-subject reliability throughout follow-up, emphasising the value of a longitudinal design and a well-characterised population, and further support the presence of a consistent longitudinal signal despite the substantial proportion of measurements below the LOD. Another limitation is that we measured the inactive circulating metabolite dp-ucMGP. To our knowledge, no assays are currently available to directly quantify the predominantly tissue-bound active form of MGP in plasma or serum. Although small amounts of the active form have been proposed to enter the circulation, this has not been firmly established. Consequently, several previous studies have similarly used circulating dp-ucMGP as a surrogate biomarker of MGP biology (12, 14, 20, 32). Potential diurnal variation in dp-ucMGP concentrations is unknown. As diurnal variation is well established for some, but not all, markers of bone turnover (42), an effect of sampling time cannot be excluded. Blood sampling was restricted to 08:00–10:00 in the healthy controls, whereas samples from girls with CPP were collected throughout a wider time interval from 08:00 to 14:00. Whether this difference in sampling time may have influenced the observed results unfortunately remains unknown. However, one study assessing total circulating MGP in healthy adults found low within-day variations with no distinct circadian pattern (43). Our interpretations are further limited by relatively small cohorts. However, data on dp-ucMGP in healthy paediatric cohorts are sparse and, to our knowledge, have not previously been assessed in girls with CPP.

Our findings provide novel insights into longitudinal changes in circulating dp-ucMGP during pubertal development and its potential association with skeletal maturation, supporting a role for MGP in growth dynamics. Future studies in larger cohorts and mechanistic approaches are warranted to further investigate the role of MGP in growth plate physiology and to explore MGP as a biomarker in children with growth disorders.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the work reported.

Funding

The Copenhagen Puberty Study was supported by the Kirsten and Freddy Johansen’s Foundation and the Danish Medical Research Council (271-05-0337). LJM received financial support from the Lundbeck Foundation and Candy’s Foundation.

Data availability

Due to sensitive clinical information, data are not publicly available. Requests for access to de-identified data can be directed to the corresponding author.