Circulating asparagine and short stature

Abstract

Short stature is phenotypically heterogeneous, and the role of circulating amino acids remains unclear. We compared plasma asparagine concentrations in children with short stature and healthy controls and examined genetically predicted circulating metabolites using two-sample Mendelian randomization (MR). We included 96 children with a clinical record diagnosis of short stature and 96 healthy controls. Plasma asparagine was measured by enzyme-linked immunosorbent assay, and the 2 groups were compared using the Mann–Whitney U test. Summary statistics for 1091 plasma metabolites and 309 metabolite ratios were screened against a European-ancestry short-stature outcome using two-sample MR. Benjamini–Hochberg false discovery rate correction was applied across 1400 inverse-variance weighted tests. Median plasma asparagine was 54.80 ng/L (interquartile range = 50.38–64.60) in children with short stature and 84.06 ng/L (71.00–95.35) in controls (P < .001). Five metabolite exposures met the nominal inverse-variance weighted threshold and sensitivity filters. Genetically predicted plasma free asparagine was inversely associated with short stature (odds ratio = 0.565; 95% confidence interval = 0.420–0.759; P = .00015; q = .106). None of the 1400 tests met q < .10. Children with short stature had lower plasma asparagine than controls. The MR estimate was in the same direction but did not survive multiple-testing correction. These data do not establish causality or support asparagine supplementation.

Key Points.

Children with short stature had lower measured plasma asparagine than healthy controls in the clinical cohort.

In two-sample Mendelian randomization, higher genetically predicted plasma free asparagine was associated with lower odds of short stature, but the association did not survive multiple-testing correction.

The findings do not establish causality or support asparagine supplementation to promote linear growth.

1. Introduction

Clinical short stature is generally defined as height more than 2 standard deviations below the population mean for age and sex and encompasses genetic, skeletal, endocrine, nutritional, and chronic-disease etiologies.^[1,2] Short stature is not synonymous with stunting, a population-level indicator of impaired linear growth commonly linked to chronic undernutrition. The 2 phenotypes may overlap, but their definitions, causes, and clinical implications differ.

Linear growth depends on growth-plate chondrogenesis, nutritional status, and the growth hormone–insulin-like growth factor 1 axis.^[2,3] Metabolomic studies have found altered amino acid and energy metabolism in children with growth hormone deficiency or idiopathic short stature, although the reported metabolites differ by age and etiologic subtype.^[3–5] Case-control data alone cannot determine whether a metabolite change is a cause or a consequence of disease, diet, or treatment.

We selected asparagine for clinical evaluation because lower circulating asparagine has been reported in children with nutritional stunting,^[6] and experimental studies show that asparagine can support amino acid exchange, protein synthesis, and mechanistic target of rapamycin complex 1 signaling.^[7,8] These experiments were conducted mainly in cellular systems and do not demonstrate a skeletal effect. The objectives were to compare plasma asparagine between children with short stature and healthy controls and to screen 1091 plasma metabolites and 309 metabolite ratios for associations with short stature using two-sample Mendelian randomization (MR).

2. Materials and methods

2.1. Participant recruitment and clinical characterization

This single-center case-control study included 96 children with a clinical record diagnosis of short stature and 96 healthy controls from Jiaxing Maternity and Children Health Care Hospital. The available analytical dataset contained group assignment and plasma asparagine concentration. Recruitment dates, age, sex, anthropometric measurements, original diagnostic work-up, diagnosing clinician, and etiologic subtype were unavailable. Consequently, we could not assess matching, adjust for these variables, conduct subgroup analyses, or prepare a participant-flow diagram. No a priori sample-size calculation was available in the study records. The study was approved by the Ethics Committee of Jiaxing Maternity and Children Health Care Hospital (approval number: 嘉妇保伦审2022研第182号; acceptance number: KY-2022-198; July 13, 2022). Written informed consent was obtained from a parent or legal guardian of each child in accordance with the approved protocol.

2.2. Plasma asparagine measurement by enzyme-linked immunosorbent assay

Peripheral venous blood was collected, and plasma was separated by centrifugation at 4000 rpm for 10 minutes. Information on fasting, collection time, anticoagulant tube, processing interval, centrifugation temperature, storage duration, and freeze–thaw cycles was unavailable. Plasma asparagine was measured with a 96-well Human Asparagine enzyme-linked immunosorbent assay (ELISA) kit (MEIMIAN [mmbio]; catalog number: MM-610063H1) according to the manufacturer’s instructions. The manual describes a double-antibody sandwich assay with a horseradish peroxidase–labeled detection antibody and tetramethylbenzidine substrate and lists a measurement range of 3 to 140 ng/L. Samples were diluted 5-fold by combining 10 μL plasma with 40 μL sample diluent. Two 30-minute incubations at 37°C were separated by a wash step. After color development and reaction termination, absorbance was read at 450 nm within 15 minutes.

Standards of 7.5, 15, 30, 60, and 120 ng/L were fitted with a second-order polynomial (R² = 0.9978). Interpolated concentrations were multiplied by the 5-fold dilution factor. The manual and available assay records did not report the lower limit of detection, lower limit of quantification, intra-assay or inter-assay coefficients of variation, recovery, dilution linearity, or cross-reactivity. The assay was not cross-validated by targeted liquid chromatography–tandem mass spectrometry. We therefore used the ELISA results for the between-group comparison but interpreted the absolute concentrations cautiously.

2.3. Two-sample MR design

We performed a two-sample MR analysis of 1091 plasma metabolites and 309 metabolite ratios against short stature.^[9] Under the instrumental-variable framework, a causal interpretation requires 3 core assumptions^[10]:

1. Relevance: the genetic variants are associated with the metabolite exposure.

2. Independence: the genetic variants are not associated with confounders of the exposure–outcome association.

3. Exclusion restriction: the genetic variants affect short stature only through metabolite exposure.

All contributing genome-wide association study (GWAS) studies reported ethical approval and participant informed consent. MR reporting followed the STROBE-MR statement.^[10]

2.4. Short-stature outcome GWAS

Summary statistics for short stature were obtained from the MRC Integrative Epidemiology Unit GWAS repository using FinnGen endpoint finn-b-E4_SHORT. The dataset included 211,416 individuals of European ancestry (293 cases and 211,123 controls) and approximately 16.38 million variants. Clinical subtype information was unavailable in the summary data, so correspondence with the Chinese clinical cohort could not be assessed.

2.5. Plasma metabolite GWAS

GWAS summary statistics for 1091 plasma metabolites and 309 metabolite ratios were obtained from Chen et al.^[9] In that study, metabolites were quantified using the Metabolon platform in 8299 unrelated participants of European ancestry; metabolite concentrations were log-transformed, outliers were removed, and traits were standardized before association testing. The complete exposure list is provided in Table S1, Supplemental Digital Content 1.

2.6. Instrument selection

For each metabolite exposure, single-nucleotide polymorphisms (SNPs) associated at P < 1 × 10⁻⁵ were clumped in TwoSampleMR using linkage disequilibrium r² < 0.001 within a 10,000-kb window.^[11–14] Instruments with F statistics < 10 were excluded. Exposure and outcome alleles were harmonized before MR analysis. R² and F statistics were calculated from the reported effect estimates, effect-allele frequencies, standard errors, and exposure sample sizes.

$${\displaystyle R^2=\frac{2\underset{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}{\overset{2}{\beta }}\mathrm{e}\mathrm{a}{\mathrm{f}}_{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}(1-\mathrm{e}\mathrm{a}{\mathrm{f}}_{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}})}{\begin{array}{c}{\displaystyle }2\underset{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}{\overset{2}{\beta }}\mathrm{e}\mathrm{a}{\mathrm{f}}_{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}(1-\mathrm{e}\mathrm{a}{\mathrm{f}}_{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}})\\ & {\displaystyle }+2\underset{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}{\overset{2}{\mathrm{s}\mathrm{e}}}\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{s}\mathrm{i}\mathrm{z}{\mathrm{e}}_{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}\mathrm{e}\mathrm{a}{\mathrm{f}}_{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}(1-\mathrm{e}\mathrm{a}{\mathrm{f}}_{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}})\end{array}}.}$$

$$\begin{array}{c}{\displaystyle }F=\frac{R^2(\mathrm{s}\mathrm{a}\mathrm{m}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{s}\mathrm{i}\mathrm{z}{\mathrm{e}}_{\mathrm{e}\mathrm{x}\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{u}\mathrm{r}\mathrm{e}}-2)}{1-R^2}.\end{array}$$

In the equations, β is the SNP–exposure effect estimate, EAF is the exposure effect-allele frequency, SE is the standard error of β, and N is the exposure GWAS sample size.

2.7. Statistical analysis

Clinical analyses were conducted in R. Distributional form was assessed using the Shapiro–Wilk test. Because plasma asparagine was non-normally distributed and variances differed between groups, values are reported as median (interquartile range) and were compared using a two-sided Mann–Whitney U test.

MR analyses used R (version 4.3.1) and TwoSampleMR. The primary estimator was inverse-variance weighted (IVW), with weighted-median, MR-Egger, weighted-mode, and simple-mode estimates used as sensitivity analyses. Cochran Q assessed heterogeneity; the MR-Egger intercept and MR-PRESSO global test assessed horizontal pleiotropy; and leave-one-out analyses assessed influential instruments. IVW odds ratios are reported per genetically predicted 1-standard-deviation higher exposure.

Multiple testing across the 1400 IVW analyses was controlled using the Benjamini–Hochberg procedure. Results with q < .05 were considered statistically significant, and results with .05 ≤ q < .10 were considered suggestive. Associations with nominal IVW P < .05, consistent directions across sensitivity estimators, and no evidence of heterogeneity or directional pleiotropy were treated as exploratory candidates; they were not false discovery rate-significant.

3. Results

3.1. Clinical cohort and plasma asparagine

The analysis included 96 healthy controls and 96 children with a clinical record diagnosis of short stature. The available dataset lacked age, sex, anthropometric measurements, original diagnostic work-up, etiologic subtype, and relevant clinical covariates. Baseline demographic statistics, adjusted analyses, and etiologic subgroup analyses could therefore not be reported.

The quadratic standard curve covered 7.5 to 120 ng/L and fit the observed optical-density values (R² = 0.9978; Fig. 1A). Plasma asparagine was lower in the short-stature group (median = 54.80 ng/L; interquartile range = 50.38–64.60) than in controls (median = 84.06 ng/L; interquartile range = 71.00–95.35; Mann–Whitney U = 1264.5; P < .001; Fig. 1B).

Figure 1.

Plasma asparagine measurement and group comparison. (A) Second-order polynomial standard curve based on 7.5, 15, 30, 60, and 120 ng/L standards; points show measured absorbance at 450 nm and the line shows the fitted curve (R² = 0.9978). (B) Plasma asparagine distributions in 96 healthy controls and 96 children with short stature. Boxes show medians and interquartile ranges, whiskers extend to 1.5 times the interquartile range, and points represent individual participants. Groups were compared using a two-sided Mann–Whitney U test (P < .001).

3.2. Exploratory metabolite-wide MR findings

After instrument selection and harmonization, 33,267 SNP–exposure instruments contributed across the 1400 metabolite and metabolite-ratio analyses (Table S2, Supplemental Digital Content 2).

Five exposures met the nominal IVW threshold and sensitivity criteria, but none survived multiple-testing correction at q < .10 (Table 1). Genetically predicted plasma free asparagine was inversely associated with short stature (odds ratio = 0.565; 95% confidence interval = 0.420–0.759; P = .000152; q = .106). The other nominal findings were X-23739, octadecadienedioate (C18:2-DC), 3-hydroxyoctanoylcarnitine (2), and the alanine-to-asparagine ratio. All 5 findings were treated as exploratory. Corresponding forest and scatter plots are shown in Figures 2 and 3, respectively.

Table 1.

Exploratory inverse-variance weighted associations with short stature.

Exposure	IVW OR (95% CI)	P value	BH q value
Octadecadienedioate (C18:2-DC)	1.551 (1.200–2.004)	.000806	.376
3-Hydroxyoctanoylcarnitine (2)	2.863 (1.677–4.889)	.000117	.106
Plasma free asparagine	0.565 (0.420–0.759)	.000152	.106
X-23739	0.538 (0.365–0.793)	.001765	.494
Alanine-to-asparagine ratio	1.665 (1.226–2.262)	.001111	.389

BH = Benjamini–Hochberg, CI = confidence interval, IVW = inverse-variance weighted, OR = odds ratio.

Figure 2.

Forest plots of exploratory nominal IVW associations between genetically predicted metabolite exposures and short stature. Points show odds ratios per 1-standard-deviation higher genetically predicted exposure, and horizontal lines show 95% CIs. Five exposures met nominal IVW and sensitivity filters; none met q < .10 after Benjamini–Hochberg correction. CI = confidence interval, IVW = inverse-variance weighted, OR = odds ratio.

Figure 3.

Scatter plots for the 5 exploratory metabolite–short-stature associations. Each point represents an instrumental SNP; the x-axis shows the SNP–exposure association and the y-axis the SNP–outcome association. Lines indicate slopes estimated using IVW, MR-Egger, weighted-median, simple-mode, and weighted-mode methods. MR = Mendelian randomization, SNP = single-nucleotide polymorphism.

Cochran Q tests did not identify statistically significant heterogeneity (Table S3, Supplemental Digital Content 3), MR-Egger intercepts did not indicate directional horizontal pleiotropy (Table S4, Supplemental Digital Content 4), and MR-PRESSO global tests did not identify significant outliers (Table S5, Supplemental Digital Content 5). No single SNP materially changed the IVW estimates in leave-one-out analyses. These leave-one-out results are shown in Figure 4.

Figure 4.

Leave-one-out analyses for the 5 exploratory metabolite–short-stature associations. Each row shows the IVW estimate recalculated after omitting 1 SNP; horizontal lines indicate 95% CIs, and the overall IVW estimate is shown for comparison. CI = confidence interval, IVW = inverse-variance weighted, SNP = single-nucleotide polymorphism.

4. Discussion

Children with short stature had lower plasma asparagine than controls. The MR estimate for genetically predicted plasma free asparagine was also inverse, but the association was nominal (q = .106), and none of the 1400 exposures met q < .10. The clinical difference and nominal MR estimate require independent replication and do not show that lower asparagine causes short stature.

Previous metabolomic studies have also reported altered amino acid metabolism in childhood growth disorders. Nuclear magnetic resonance studies found distinct metabolic profiles in growth hormone deficiency and idiopathic short stature,^[4,5] and a serum-metabolomics study reported subtype-specific changes in childhood short stature.^[3] Lower circulating asparagine was observed in stunted Malawian children,^[6] although nutritional stunting is not equivalent to clinical short stature. Differences in age, nutrition, and the cause of short stature may contribute to variation across studies.

Asparagine contributes to protein synthesis and cellular amino acid homeostasis. In cell systems, intracellular asparagine can exchange with extracellular amino acids and coordinate protein and nucleotide synthesis,^[7] and it can activate mechanistic target of rapamycin complex 1 through an ADP-ribosylation factor 1–dependent pathway.^[8] Mechanistic target of rapamycin complex 1 links nutrient availability to anabolic growth. These findings suggest a possible mechanism, but the experiments were not conducted in pediatric growth plates. Whether physiological differences in circulating asparagine affect chondrocyte proliferation or human linear growth is unknown.

The clinical association may be confounded by dietary protein intake, nutritional status, pubertal stage, liver and renal function, endocrine disorders, chronic inflammation, medication use, and etiologic subtype. Data on these factors were unavailable, preventing matching, covariate adjustment, and subgroup analysis. A single measurement also does not capture temporal variation, and ELISA is less analytically specific than targeted liquid chromatography–tandem mass spectrometry.

The clinical cohort was small and came from a single center. For the MR analysis, selecting instruments at P < 1 × 10⁻⁵ carries more risk of weak-instrument bias than using genome-wide significant variants, although SNPs with F < 10 were excluded. Nonsignificant heterogeneity, MR-Egger, and MR-PRESSO tests do not rule out balanced or pathway-specific pleiotropy. Lifelong genetically predicted differences may also differ from dietary or pharmacologic changes in asparagine.^[10] The outcome GWAS included only 293 cases, and none of the 5 nominal associations survived multiple-testing correction. Differences in ancestry, environment, phenotype definition, and etiologic composition between the Chinese clinical cohort and the European-ancestry GWAS further limit generalizability.

The present data do not justify asparagine supplementation. No intervention, dose-response, growth-velocity, or safety data were collected, and a genetic association cannot be assumed to represent a dietary or pharmacologic effect. Replication should use a larger, phenotypically characterized pediatric cohort and standardized fasting samples. Future validation studies should preferably use standardized targeted metabolomic platforms, particularly targeted liquid chromatography–tandem mass spectrometry, to improve analytical validity and facilitate comparisons across studies. MR replication will require stronger instruments and a larger outcome GWAS. Mechanistic or interventional studies would be appropriate only after the associations are reproduced.

5. Conclusions

Children with short stature had lower plasma asparagine than healthy controls. The MR estimate for genetically predicted plasma free asparagine was inverse but did not survive multiple-testing correction. These findings warrant replication but do not establish asparagine as a causal factor or therapeutic target.

Acknowledgments

This work was supported by the Medical Health Science and Technology Project of the Zhejiang Provincial Health Commission (grant number 2023KY337) and the Public Welfare Research Project of Jiaxing (2025CGW059).

Author contributions

Conceptualization: Jianguo Wang.

Data curation: Siyi Zhang.

Writing – original draft: Dongdong Chen, Keng Ling.

Writing – review & editing: Huan Zhou, Jianguo Wang.