Insights into the pathophysiology of catch-up compared with non-catch-up growth in children born small for gestational age: an integrated analysis of metabolic and transcriptomic data

Article metrics


Small for gestational age (SGA) children exhibiting catch-up (CU) growth have a greater risk of cardiometabolic diseases in later life compared with non-catch-up (NCU) SGA children. The aim of this study was to establish differences in metabolism and gene expression profiles between CU and NCU at age 4–9 years. CU children (n=22) had greater height, weight and body mass index standard deviation scores along with insulin-like growth factor-I (IGF-I) and fasting glucose levels but lower adiponectin values than NCU children (n=11; all P<0.05). Metabolic profiling demonstrated a fourfold decrease of urine myo-inositol in CU compared with NCU (P<0.05). There were 1558 genes differentially expressed in peripheral blood mononuclear cells between the groups (P<0.05). Integrated analysis of data identified myo-inositol related to gene clusters associated with an increase in insulin, growth factor and IGF-I signalling in CU children (P<0.05). Metabolic and transcriptomic profiles in CU SGA children showed changes that may relate to cardiometabolic risk.


Children born small for gestational age (SGA) have an increased risk of mortality and morbidity in the neonatal period,1 higher rates of learning disability2, 3, 4, 5 and a greater risk of a range of diseases later in life, including cardiometabolic6, 7, 8 and renal conditions.9

SGA is defined as a weight or length at birth lesser than −2 standard deviation scores below the mean for gestational age;10 it may be associated with fetal growth restriction (FGR) where genetic growth potential is not reached due to an insult occurring in utero.11 In the early years of life, most children born SGA exhibit catch-up (CU) growth. However, 10% (approximately 2000/year in the United Kingdom) do not catch-up (non-catch-up; NCU) and are eligible for growth hormone treatment from the age of 4 years in Europe.10, 12

Several reports have shown that children born SGA exhibiting CU growth have a greater risk of developing cardiovascular and metabolic disease in later life compared with those who do not CU.8, 13, 14, 15 The metabolic disorders include dyslipidaemia, insulin resistance and type 2 diabetes, and can be detected from early childhood.16, 17 The underlying mechanisms generating the different metabolic profiles and growth patterns between CU and NCU SGA children are not well understood.

Metabolic perturbations in children born SGA are expected and targeted investigations of individual serum metabolites have demonstrated this.18, 19, 20 The untargeted measurement of endogenous and exogenous metabolites, defined as metabolomics,21 is a rapidly developing tool to holistically study metabolism. Metabolomics has been used to assess metabolic profiles in urine from FGR neonates compared with controls to define the metabolic patterns associated with this condition22 and also in placental explants from the mothers of SGA children23 and peripheral plasma samples from early pregnancies later shown to be associated with the birth of an SGA child.24 Metabolomic studies have not yet been performed directly in SGA children in childhood.

A potential role for phospholipids and their associated biological actions has been identified in SGA. They have been shown to have potential predictive value in the identification of SGA during early pregnancy24 and an increase of myo-inositol, an essential precursor for the production of inositol phospholipids, has been shown in FGR neonates,22 suggesting that it could potentially be associated with the predisposition to type 2 diabetes and metabolic diseases in adulthood. In contrast, several studies have found that myo-inositol has an insulin-promoting effect, by increasing glucose cellular transport and by modulating the intracellular insulin signalling.25, 26 Myo-inositol is a component of structural and signalling lipids such as phosphatidylinositol (PI) and phosphatidylinositol phosphate. Inositol derivatives are associated with insulin resistance.27

Analysis of gene expression has been used to investigate SGA. Using this approach revealed lower expression of genes from the growth hormone locus28 and reduced expression of the insulin-like growth factor -2 (IGF-2) gene29 associated with SGA. Whole-genome analysis of gene expression, termed transcriptomics, has been used to investigate placentas from SGA pregnancies and demonstrated an increased antiangiogenic gene expression signature in comparison with controls.30 The use of transcriptomics to compare cord blood from SGA babies with controls has been inconclusive.31 Transcriptomic studies have not been undertaken in older SGA children, once the pattern of growth has been established (CU versus NCU).

Progress in the study of large biological data sets has resulted in the development of integrated approaches to the analysis of combined transcriptomic and metabolic data sets. These methodologies have been facilitated by the use of network biology to place observations in the context of the interactome,32 a model of all biomolecular interactions, and thereby to increase confidence in the findings.33 An integrated, multi-omic approach reduces the noise in statistical interpretation.34 Also network biology is an effective approach to study small data sets as it focuses on the importance of system units (‘network clusters’)35, 36, 37 as a measure of similarity. This approach is also robust to random variation in the data,38 along with differences in network sizes (‘scale-free’ property).39

This is the first report integrating metabolic and transcriptomic profiles from prepubertal children born SGA aimed at comparing differences between children who have exhibited CU growth with those who did not CU. Furthermore, we have integrated these different approaches using network analysis33, 40 to identify putative transcriptomic and metabolic changes related to growth and metabolic patterns in SGA children (Figure 1).

Figure 1

Schema of study design. The differences between small for gestational age children with catch-up (CU) and without catch-up (NCU) growth were examined by studying metabolic and gene expression changes. Transcriptomic and metabolomic data sets were then integrated using network biology to assess function.

PowerPoint slide

Materials and methods

Study design

All children recruited were aged between 4 and 9 years of age, and presented a birth weight lesser than −2 standard deviation scores for gestational age. Participants were stratified into two groups based on growth pattern: CU and NCU growth. The former was defined as occurring when an SGA infant had crossed over the third percentile for height with a growth velocity greater than the median for chronological age and gender.41 Approximately 100 families were approached, and 22 CU and 11 NCU children attended clinic appointments for research assessment and for collection of biological samples (Table 1). Different subgroups of this cohort were available for subsequent analysis (Figure 1 and Supplementary Table S1).

Table 1 Cohort demographics and auxology of all SGA children (n=33)

Measurement of serum biomarkers

Serum biomarkers were measured by enzyme-linked immunosorbent assay according to the manufacturer’s protocol. Serum IGF-I levels were measured using the Beckman Coulter kit (Beckman Coulter, High Wycombe, UK) and converted to standard deviation scores according to reference values of age and gender.42 Plasma insulin and glucose levels were determined by using kits from Mercodia (Mercodia, Sweden). Adiponectin was measured using the R&D Systems kit (Abingdon, UK). Insulin resistance was assessed by using the homoeostasis model assessment of insulin resistance, defined by (fasting insulin (mU l−1) × fasting glucose (mmol l−1)/22.5).43

Metabolic profiling

Metabolomic analysis of serum samples (Supplementary Table S1) was performed by applying Ultra Performance Liquid Chromatography-Mass Spectrometry (UPLC-MS; Acquity UPLC (Waters, Elstree, UK) coupled to a hybrid LTQ-Orbitrap XL mass spectrometer (Hemel Hempstead, UK) and Gas Chromatography-Mass Spectrometry (GC–MS; 6890 GC (Agilent Technologies, Cheadle, UK) and Pegasus III mass spectrometer (Leco, Stockport, UK)) as described previously and included the use of quality control samples.44 Metabolomic analysis of urine samples (Supplementary Table S1) was performed using GC–MS (6890 GC (Agilent Technologies) and Pegasus III mass spectrometer (Leco)). Further details are provided in the Supplementary Information.


Peripheral blood mononuclear cells have been demonstrated to be an appropriate model system to study growth response.45, 46, 47 Total RNA was prepared from peripheral blood mononuclear cells samples (Supplementary Table S1) using the RNeasy kit (Qiagen GmbH, Hilden, Germany). Complementary RNA was generated by Affymetrix gene profiling reagents (Affymetrix, Santa Clara, CA, USA). Arrays were then scanned on an Affymetrix GeneChip 3000 7G scanner. Further details are available in the Supplementary Information.

Statistical analysis

Suitable metabolic and transcriptomic blood samples were available on subsets of the cohort (Supplementary Table S1). Three outputs were evaluated: auxological and biochemical data, metabolomic and transcriptomic data.

Auxological and biochemical data were expressed as median and interquartile ranges (median (Q1,Q3)). Differences in gender and ethnicity prevalence between groups were assessed by χ2-test, differences in other variables were assessed by the Mann–Whitney U-test. Statistical analysis was performed using SPSS programme (Statistical Package for Social Science), version 20.0 software for Windows (SPSS, Chicago, IL, USA).

Multivariate principal component analysis of the metabolomic data was performed to assess variability associated with analysis order and class. Univariate analysis was performed using the Wilcoxon signed-rank test.

For transcriptomic data, analysis of variance was used to compare gene expression changes between the groups, and cross-validation was performed by principal component analysis (Qlucore Omics Explorer 2.2, Qlucore, Lund, Sweden). Enriched gene functions and canonical biological pathways were identified with ingenuity pathways analysis software (Ingenuity Systems, Redwood City, CA, USA) and by logistic regression using LRPath48 using a Benjamini-Hochberg correction for false discovery rate.49

P-values <0.05 or false discovery rate modified P-values (q) <0.2 were considered statistically significant. Detailed information is available in the Supplementary Information.

Network analysis

Three network analysis approaches were used, as follows:

  1. 1)

    Metabolic profiles were mapped to pathways associated with gene expression changes using ingenuity pathways analysis software.50

  2. 2)

    Gene expression data were mapped onto the inferred interactome networks generated from metabolomic profiles using ingenuity pathways analysis.50

  3. 3)

    Generation and integration of metabolic and gene expression profiles was performed using the Metscape plugin for Cytoscape To identify associated pathways, logistic regression of all metabolism-related gene expression was performed using LRPath.48


Auxological data

The patient group consisted of 33 prepubertal children born SGA (Table 1). Twenty-two children exhibited CU growth and 11 NCU growth with no difference in the age range (P=0.133) or ethnic background (P=0.354). Prepubertal children with CU were significantly taller and heavier with a greater body mass index compared with NCU children (Table 1).

Measurement of serum biomarkers

The SGA patients who exhibited CU growth had higher levels of serum IGF-I compared with NCU children (142.0 (81.0, 188.5) vs 74.5 (43.5, 90.7) μg l−1, P=0.007)). Children with CU had similar insulin and homoeostasis model assessment of insulin resistance compared with NCU (insulin: 3.5 (0.5, 3.6) vs 2.6 (0.8, 4.8) mU l−1, P=0.857 and homoeostasis model assessment of insulin resistance: 1.5 (1.0, 1.6) vs 0.7 (0.6, 1.0), P=0.071). However, fasting glucose was significantly increased in the CU group compared with the NCU group (5.8 (5.0, 5.9) vs 3.8 (3.7, 4.7) mmol l−1, P=0.032) and their adiponectin levels were lower (2.9 (1.6, 4.1) vs 10.4 (5.3, 15.4) mg l−1, P=0.016; Table 1).

Metabolic profiles

Subsets of the cohort recruited were used for metabolic profiling (Supplementary Table S1). Principal component analysis of data acquired for serum and urine showed no visible separation of individual data based on analysis order or class (data not shown). Univariate analysis showed that five metabolites had significant differences between SGA CU and NCU children by GC–MS in serum or urine samples (Table 2): myo-inositol (in urine) and decanoic acid (in serum) were decreased by 4-fold and 1.6-fold, respectively, in SGA children with CU in comparison with NCU. The other metabolites detected (glutamine (in serum), uric acid and carnitine (in urine)) were significantly increased in CU compared with NCU (Table 2). In addition, UPLC-MS detected 68 metabolites with significant differences between CU and NCU in serum; these metabolites can be separated into classes including glycerophospholipids, fatty acids, nucleotide metabolites and bile acids (Supplementary Table S2).

Table 2 Metabolites measured by gas chromatography-mass spectrometry in serum and urine samples

Gene expression profiles

Principal component analysis identified one outlier by cross-validation and that was removed from subsequent analysis; no further separation of the data was observed (data not shown). The remaining three CU were compared with the five NCU children using analysis of variance and 2061 gene probe-sets (from 1558 unique genes—Supplementary Table S3) were identified as differentially expressed between the groups (P<0.05).

The 1558 genes with differential expression between CU and NCU children were associated with growth, metabolism, molecular and cellular functions (Supplementary Table S4). Genes that were differentially expressed between CU and NCU SGA children clustered in growth factor pathways, including vascular endothelial growth factor (P<0.05). Metabolic pathways with gene expression changes included the synthesis of myo-inositol derivatives (P<0.05). Intracellular signalling pathways were also associated with differentially expressed genes, including G-protein signalling, cAMP signalling, protein kinase A signalling and sphingosine-1-phosphate signalling (P<0.05; Supplementary Table S5). Of note, sphingosine-1-phosphate was also identified as significant in the UPLC-MS metabolomics data set (Supplementary Table S2).

A logistic regression analysis of the biological pathways associated with the gene expression differences between CU and NCU was performed and 134 pathways were defined as significantly directionally altered (q<0.2). This analysis identified that growth factor pathways (IGF-I, EGF, FGF, VEGF and PDGF) and PI 3-kinase signalling were all upregulated in the CU group (Supplementary Table S6).

Network analysis approaches

In order to integrate all metabolomic and transcriptomic data via network analysis and to extrapolate pathways with biological function, three different approaches were adopted.

  1. 1)

    Mapping of metabolite profiles to gene expression associated pathways: Metabolites identified by GC–MS and UPLC-MS as different between CU and NCU were mapped onto the pathways associated with gene expression differences (Supplementary Tables S5 and S6). This approach revealed a correlation in the CU group between decreased myo-inositol and increased 1-phosphatidyl-1D-myo-inositol 3-phosphate and PI metabolite levels (Supplementary Figure S1). Alteration in gene expression was also associated with the observed metabolic changes in this pathway within two enzyme complexes (Supplementary Figure S1): PI 3-kinase (Enzyme commission number (EC), which was associated with increased expression of PTPN11 (P<0.01, Supplementary Table S3), and PI-4-phosphate 3-kinase (EC2.7.1.154) associated with decreased expression of PIK3C3 and PIK3C2A (P<0.05, Supplementary Table S3).

  2. 2)

    Gene expression data mapped to inferred metabolic network: The metabolic differences identified between the CU and NCU groups were used independently to infer networks using ingenuity pathways analysis.50 This procedure identified one network containing four of the metabolites (uric acid, carnitine, decanoic acid and glutamine), which was associated with molecular transport (P<1 × 10−12; Figure 2). Changes in gene expression were mapped to this network (upregulated: ATF6, NQO1 part of the super oxide dismutase complex, ACADL and downregulated: ASNS, IL23A, ACTA2 part of the Alpha actin complex); associated biological pathways were shown to include nucleotide metabolism (P<9.0 × 10−4), vascular endothelial growth factor signalling (P<9.8 × 10−4), RHO signalling (P<1.8 × 10−3), mitochondrial function (P<4.6 × 10−3) and peroxisome proliferator-activated receptor/retinoid X receptor gamma activation (P<5.6 × 10−3). Myo-inositol was also linked to 17 gene expression changes in the data set that were related to Wnt/Notch pathway signalling (P<2.7 × 10−5), a growth pathway also shown to be upregulated in the CU group by logistic regression (Supplementary Table S6).

    Figure 2

    Gene expression data mapped onto the inferred networks derived from metabolic profiles: Integration of metabolic and gene expression changes by functional inference (IPA). Green, decreased expression; red, increased expression.

    PowerPoint slide

  3. 3)

    Generation of an inferred networks derived from the integration of metabolite and gene expression profiles: Metabolites detected by GC–MS and UPLC-MS as statistically different between CU and NCU groups were combined with the gene expression differences in the metabolic pathways defined by logistic regression (q<0.2); this data set was used to infer an integrated network of 1249 nodes (807 proteins and 442 metabolites). A network cluster was observed surrounding the myo-inositol metabolite (Figure 3a), related to an upregulation of adrenaline/noradrenaline synthesis and insulin signalling pathways (q<0.02) (Figure 3a). The phosphatidyl inositol network submodule containing myo-inositol (dotted red line, Figure 3b) was associated with decreased expression of GLO1 and AKR1B1 genes (Supplementary Table S3), which have roles in antiapoptosis and cell growth, respectively, along with increased expression of the GCK gene that has a role in glycolysis and gluconeogenesis (Figure 3b).

    Figure 3

    Integration of metabolite and gene expression profiles into an inferred network. Network analysis using MetScape. (a) An integrated network cluster surrounding the myo-inositol metabolite derived from transcriptomic (dark blue circles, protein derived from ‘seed’ gene; light blue circles, inferred protein interaction) and metabolomic differences (red hexagon, seed metabolite; pink hexagon, inferred metabolite) between CU and NCU SGA. Associated biological pathways were determined using logistic regression (LRPath, red, pathways with increased action; green, pathways with decreased action). Dotted red line, position of the phosphatidyl inositol subcluster containing myo-inositol. (b) The phosphatidyl inositol submodule from the integrated network, showing the relationship of myo-inositol to gene expression changes associated with metabolic action; green arrows, decreased expression; red arrows, increased expression. CU, catch-up; NCU, non-catch-up; SGA, small for gestational age.

    PowerPoint slide


This is the first study in SGA children that has used metabolomics as an approach to investigate changes in metabolites and combined these data with transcriptomic data using network analysis. This has allowed us to take a multi-dimensional ‘systems’ view to show that in the early childhood years, metabolic and gene expression profiles are markedly different in CU compared with NCU SGA children. Potential new markers as well as insights into mechanism for divergent postnatal growth patterns in SGA children and their associated later life metabolic risks have been identified.

By definition, the prepubertal children with CU growth were taller and heavier and had greater body mass index compared with NCU. The single analyte serum markers showed higher IGF-I levels in CU growth children consistent with their growth pattern, but lower adiponectin levels, as recently described, and potentially consistent with an early indicator of cardiometabolic risk.52, 53 Adiponectin is a protein derived from the adipose tissue in humans and has been related to obesity and insulin resistance.54 There is functional evidence that it has a protective role in the development of metabolic complications, as high concentrations have been associated with a reduced risk of type 2 diabetes, and low levels with the converse.55 Moreover, administration of adiponectin causes glucose lowering effects and ameliorates insulin resistance in mice models.56, 57 The magnitude of difference in adiponectin levels between CU and NCU groups is much greater than the expected changes based on reference values for age and body mass index.53, 58, 59 Insulin levels and homoeostasis model assessment of insulin resistance at this stage were similar but glucose was increased in CU children in alignment with decreased adiponectin levels.56, 57 This limited biochemical analysis provides metabolic markers but no insight into cellular mechanism.

In contrast, metabolomic analysis showed, as a primary observation of this study, that the metabolite myo-inositol, which was decreased by fourfold in the urine from SGA children with CU compared with NCU, was associated with CU growth. Myo-inositol is present at micromolar concentrations in urine and blood60 and forms the structural basis for the inositol phosphate secondary messenger system involved in multiple cell functions, including insulin receptor signalling and growth pathways.26

The role of myo-inositol in insulin signalling is still controversial. Recent reports have found this metabolite to be increased in the urine of FGR neonates compared with normal birth weight children22 and the same results have been shown in serum using animal models such as in piglets.61 It was hypothesized that the altered glucose metabolism associated with FGR, shown by the increase in plasma myo-inositol, may predispose toward type 2 diabetes in adulthood.18 In contrast, several in vitro and in vivo reports have found myo-inositol to have an insulin-like effect.25, 62 It has been shown to decrease hyperglycaemia and hyperlipidaemia in rats25 and to improve insulin resistance in patients with gestational diabetes.63

In order to better characterise the role of myo-inositol, we integrated metabolomic and transcriptomic data using network analysis, differentiating CU from NCU SGA children (Figure 4). Specifically, we showed that myo-inositol metabolism was associated with the PI pathway. We demonstrated that in CU SGA children, PI 3-kinase was upregulated and that PI-4-phosphate 3-kinase was downregulated. We then investigated the activity downstream of myo-inositol in the PI signalling pathway using the metabolic and transcriptomic changes that mapped to this pathway. The upregulation of PI 3-kinase signalling was associated with increased activity of growth factor pathways (for example, IGF-I and EGF pathways) and the increase of serum IGF-I observed in CU SGA children (Figure 4). The observed downregulation of PI4PK3 in CU SGA children has been previously associated with a reduction in cell surface glucose receptors,64 which may be related to the increased serum glucose in this group52 (Table 1). We also predicted that the increased activity downstream of the PI pathways would be associated with an increase in calcium signalling, which is required for multiple cell functions including insulin secretion65 (Figure 4). Calcium channels on the pancreatic β-cell are important to stimulate fusion of the insulin vesicles to the cell membrane and to increase insulin secretion outside the blood stream via the PI pathway.65 Moreover, myo-inositol has been associated with adiponectin levels, as shown by a recent study in which the administration of this metabolite to patients with gestational diabetes significantly improved adiponectin levels with subsequent improvement of insulin resistance status.63

Figure 4

Predicted differences in activity of phosphatidyl inositol signalling in CU compared with NCU SGA. The metabolic and gene expression changes between CU and NCU growth shown in Supplementary Figure S1 are predicted to cause further changes in pathway activity (light green, deceased activity; pink, increased activity). The change in predicted activity (bold red and green lines) correlated with several gene expression changes (pathway association and logistic regression) (blue text)); observations from the literature (italics) and change in serum biomarkers (normal text).

PowerPoint slide

We also showed that myo-inositol was directly involved in insulin signalling via the Wnt pathway and has been linked to the glucokinase gene (GCK), which has a role in glycolysis and gluconeogenesis. Overall myo-inositol was associated with upregulation of insulin signalling, suggesting that the decreased myo-inositol levels in CU children might be associated with impairment in insulin signalling and be a potential early marker for the development of insulin resistance later in life.

In addition, metabolomic analysis detected other metabolites with UPLC-MS and GC–MS, such as glutamine, uric acid and carnitine, which were significantly increased in CU compared with NCU. These metabolites are associated with anabolic functions in the organism including lipid metabolism,66 nucleic acid metabolism,67 energy utilization67 and protein synthesis.68 In our study, they were associated with several biological pathways, mostly involved in growth and cellular proliferation (nucleotide metabolism, vascular endothelial growth factor and RHO signalling), in carbohydrate, lipid and protein metabolism (mitochondrial function and peroxisome proliferator-activated receptor/retinoid X receptor gamma activation) and in endocytotic processes. These observations again support the increased anabolic status in CU compared with NCU, with increased growth but also increased metabolic activity.

Furthermore, the metabolomic analysis highlighted the presence of some biological components with significant differences between CU and NCU in serum, including glycerophospholipids, fatty acids, nucleotide metabolites and bile acids (Supplementary Table S2). Particularly, bile acids were found to be decreased in CU SGA children compared with NCU. Synthesis of bile acids is known to be regulated by insulin and glucose to facilitate the absorption of nutrients from the gut.69 The metabolic signature of decreased bile acid activity in CU SGA may be, in part, due to differences in insulin signalling between CU and NCU SGA: this may involve the downregulation of expression of the alpha-methylacyl-CoA racemase enzyme (AMACR) (Supplementary Table S3), a gene essential for bile acid synthesis, as defects in AMACR have been associated with serum bile acid levels.70

Gene expression changes in peripheral blood mononuclear cells have been established as an appropriate model to assess human growth45, 46, 47 and, although not a substitute for growth plate tissue, they are growth hormone responsive.45, 47, 71 Growth is a multi-faceted process involving organism-wide changes as well as active bone growth. A key criterion proposed by Powell et al.72 for the study of function in unrelated tissues is the establishment of an overlap of transcriptional control. There is an extensive overlap between the regulation of lymphoid cell function and classic growth pathways,73, 74, 75 including key transcription factors in the proliferation and differentiation of B cells (for example, early B cell factor), which have been associated with bone development and growth.73

A limitation of this study is the lack of matched controls not born SGA. However, comparison of the SGA gene expression data with the recent observation of age-related gene expression in a data set of normal children46 showed that only 4.7% of the genes that change between CU and NCU children have an age-related component in normal development and none of these featured in the integrated analysis. Further limitations include small sample size in relation to the known confounding effects of age, ethnicity and gender along with the absence of measurements of body composition beyond body mass index. However, the integrated analysis of multiple-omic data using network analysis provides an increase in the confidence of the biological importance of the mechanisms described.33, 34 Moreover, the direct comparison of CU with NCU children allows the exploration of differences specifically within the SGA phenotype with emphasis on understanding pathophysiological mechanisms. The age band used in this study ranged between 4 and 8 years, and this does not result in any distinct cluster of age-related gene expression as previously shown in normal children.46 Moreover, the CU and NCU groups remain distinct clinical entities as it is highly unlikely that the NCU group would start to show CU between 4 and 8 years of age, as CU growth in SGA usually occurs during the postnatal period within the first 2 years of life.10, 76, 77 Larger prospective studies across a broader age range (including early infancy) are required to better characterize the role of myo-inositol. The evidence of altered levels of this metabolite in early childhood may become helpful as a screening test to evaluate and monitor CU growth over a period of time, as it is easy to assess in urine samples. In addition, recognition of all the pathways related to myo-inositol change provides an opportunity to look for drug targets that could be used to prevent the development of metabolic diseases in adult life.

In conclusion, this exploratory study has provided novel mechanistic insight into the metabolic and growth patterns of SGA patients. Myo-inositol has been identified as a marker that could aid identification of CU growth and detect those SGA children where potential interventions in early life could decrease long-term cardiometabolic risk.


  1. 1

    Verlijsdonk JW, Winkens B, Boers K, Scherjon S, Roumen F . Suspected versus non-suspected small-for-gestational age fetuses at term: perinatal outcomes. J Matern Fetal Neonatal Med 2012; 25: 938–943.

  2. 2

    Pueyo V, Oros D, Valle S, Tuquet H, Güerri N, Argüelles M et al. Axonal loss and cognitive deficits in term infants with normal umbilical artery Doppler born small-for-gestational age. Ultrasound Obstet Gynecol 2012; 40: 297–303.

  3. 3

    De Bie HM, Oostrom KJ, Boersma M, Veltman DJ, Barkhof F, Delemarre-van de Waal HA et al. Global and regional differences in brain anatomy of young children born small for gestational age. PLoS ONE 2011; 6: e24116.

  4. 4

    Puga B, Gil P, de Arriba A, Labarta JI, Romo A, Mayayo E et al. Neurocognitive development of children born small for gestational age (SGA). An update. Pediatr Endocrinol Rev 2012; 9: 716–726.

  5. 5

    Lee PA, Houk CP . Cognitive and psychosocial development concerns in children born small for gestational age. Pediatr Endocrinol Rev 2012; 10: 209–216.

  6. 6

    Barker DJ . Fetal nutrition and cardiovascular disease in later life. Br Med Bull 1997; 53: 96–108.

  7. 7

    Nobili V, Alisi A, Panera N, Agostoni C . Low birth weight and catch-up-growth associated with metabolic syndrome: a ten year systematic review. Pediatr Endocrinol Rev 2008; 6: 241–247.

  8. 8

    de Arriba A, Dominguez M, Labarta JI, Domínguez M, Puga B, Mayayo E et al. Metabolic syndrome and endothelial dysfunction in a population born small for gestational age relationship to growth and Gh therapy. Pediatr Endocrinol Rev 2013; 10: 297–307.

  9. 9

    Argente J, Mehls O, Barrios V . Growth and body composition in very young SGA children. Pediatr Nephrol 2010; 25: 679–685.

  10. 10

    Clayton PE, Cianfarani S, Czernichow P, Johannsson G, Rapaport R, Rogol A . Management of the child born small for gestational age through to adulthood: a consensus statement of the International Societies of Pediatric Endocrinology and the Growth Hormone Research Society. J Clin Endocrinol Metab 2007; 92: 804–810.

  11. 11

    Bamberg C, Kalache KD . Prenatal diagnosis of fetal growth restriction. Semin Fetal Neonatal Med 2004; 9: 387–394.

  12. 12

    Chatelain P, Colle M, Nako JP, Le Luyer B, Wagner K, Berlier P et al. Optimization of growth hormone dosing in children born small for gestational age: an open-label, randomized study of children during the fourth year of therapy. Horm Res Paediatr 2012; 77: 156–163.

  13. 13

    Vaag A . Low birth weight and early weight gain in the metabolic syndrome: consequences for infant nutrition. Int J Gynaecol Obstet 2009; 104 (Suppl 1): S32–S34.

  14. 14

    Levy-Marchal C, Czernichow P . Small for gestational age and the metabolic syndrome: which mechanism is suggested by epidemiological and clinical studies? Horm Res 2006; 65 (Suppl 3): 123–130.

  15. 15

    Leunissen RW, Kerkhof GF, Stijnen T, Hokken-Koelega AC . Effect of birth size and catch-up growth on adult blood pressure and carotid intima-media thickness. Horm Res Paediatr 2012; 77: 394–401.

  16. 16

    Ibanez L, Suarez L, Lopez-Bermejo A, Diaz M, Valls C, de Zegher F . Early development of visceral fat excess after spontaneous catch-up growth in children with low birth weight. J Clin Endocrinol Metab 2008; 93: 925–928.

  17. 17

    Veening MA, Van Weissenbruch MM, Delemarre-van de Waal HA . Glucose tolerance, insulin sensitivity, and insulin secretion in children born small for gestational age. J Clin Endocrinol Metab 2002; 87: 4657–4661.

  18. 18

    Cianfarani S, Maiorana A, Geremia C, Scire G, Spadoni GL, Germani D . Blood glucose concentrations are reduced in children born small for gestational age (SGA), and thyroid-stimulating hormone levels are increased in SGA with blunted postnatal catch-up growth. J Clin Endocrinol Metab 2003; 88: 2699–2705.

  19. 19

    Ghirri P, Ladaki C, Bartoli A, Scirè G, Vuerich M, Spadoni GL et al. Low birth weight for gestational age associates with reduced glucose concentrations at birth, infancy and childhood. Horm Res 2007; 67: 123–131.

  20. 20

    Tenhola S, Todorova B, Jaaskelainen J, Janne OA, Raivio T, Voutilainen R . Serum glucocorticoids and adiponectin associate with insulin resistance in children born small for gestational age. Eur J Endocrinol 2010; 162: 551–557.

  21. 21

    Dunn WB, Broadhurst DI, Atherton HJ, Goodacre R, Griffin JL . Systems level studies of mammalian metabolomes: the roles of mass spectrometry and nuclear magnetic resonance spectroscopy. Chem Soc Rev 2011; 40: 387–426.

  22. 22

    Dessi A, Atzori L, Noto A, Visser GH, Gazzolo D, Zanardo V et al. Metabolomics in newborns with intrauterine growth retardation (IUGR): urine reveals markers of metabolic syndrome. J Matern Fetal Neonatal Med 2011; 24 (Suppl 2): 35–39.

  23. 23

    Horgan RP, Broadhurst DI, Dunn WB, Brown M, Heazell AE, Kell DB et al. Changes in the metabolic footprint of placental explant-conditioned medium cultured in different oxygen tensions from placentas of small for gestational age and normal pregnancies. Placenta 2010; 31: 893–901.

  24. 24

    Horgan RP, Broadhurst DI, Walsh SK, Brown M, Heazell AE, Kell DB et al. Metabolic profiling uncovers a phenotypic signature of small for gestational age in early pregnancy. J Proteome Res 2011; 10: 3660–3673.

  25. 25

    Nascimento NR, Lessa LM, Kerntopf MR, Sousa CM, Alves RS, Queiroz MG et al. Inositols prevent and reverse endothelial dysfunction in diabetic rat and rabbit vasculature metabolically and by scavenging superoxide. Proc Natl Acad Sci USA 2006; 103: 218–223.

  26. 26

    Jones DR, Varela-Nieto I . The role of glycosyl-phosphatidylinositol in signal transduction. Int J Biochem Cell Biol 1998; 30: 313–326.

  27. 27

    Larner J . D-chiro-inositol—its functional role in insulin action and its deficit in insulin resistance. Int J Exp Diabetes Res 2002; 3: 47–60.

  28. 28

    Mannik J, Vaas P, Rull K, Teesalu P, Rebane T, Laan M . Differential expression profile of growth hormone/chorionic somatomammotropin genes in placenta of small- and large-for-gestational-age newborns. J Clin Endocrinol Metab 2010; 95: 2433–2442.

  29. 29

    Guo L, Choufani S, Ferreira J, Smith A, Chitayat D, Shuman C et al. Altered gene expression and methylation of the human chromosome 11 imprinted region in small for gestational age (SGA) placentae. Dev Biol 2008; 320: 79–91.

  30. 30

    Toft JH, Lian IA, Tarca AL, Erez O, Espinoza J, Eide IP et al. Whole-genome microarray and targeted analysis of angiogenesis-regulating gene expression (ENG, FLT1, VEGF, PlGF) in placentas from pre-eclamptic and small-for-gestational-age pregnancies. J Matern Fetal Neonatal Med 2008; 21: 267–273.

  31. 31

    Adkins RM, Tylavsky FA, Krushkal J . Newborn umbilical cord blood DNA methylation and gene expression levels exhibit limited association with birth weight. Chem Biodivers 2012; 9: 888–899.

  32. 32

    Stark C, Breitkreutz BJ, Chatr-Aryamontri A, Boucher L, Oughtred R, Livstone MS et al. The BioGRID Interaction Database: 2011 update. Nucleic Acids Res 2011; 39 (Database issue): D698–D704.

  33. 33

    Choi H, Pavelka N . When one and one gives more than two: challenges and opportunities of integrative omics. Front Genet 2011; 2: 105 105.

  34. 34

    Ideker T, Dutkowski J, Hood L . Boosting signal-to-noise in complex biology: prior knowledge is power. Cell 2011; 144: 860–863.

  35. 35

    Aloy P, Russell RB . Taking the mystery out of biological networks. EMBO Rep 2004; 5: 349–350.

  36. 36

    Arrell DK, Terzic A . Interpreting networks in systems biology. Clin Pharmacol Ther 2013; 93: 389–392.

  37. 37

    Barabasi AL . Scale-free networks: a decade and beyond. Science 2009; 325: 412–413.

  38. 38

    Albert R, Jeong H, Barabasi AL . Error and attack tolerance of complex networks. Nature 2000; 406: 378–382.

  39. 39

    Barabasi AL, Albert R . Emergence of scaling in random networks. Science 1999; 286: 509–512.

  40. 40

    Karnovsky A, Weymouth T, Hull T, Tarcea VG, Scardoni G, Laudanna C et al. Metscape 2 bioinformatics tool for the analysis and visualization of metabolomics and gene expression data. Bioinformatics 2012; 28: 373–380.

  41. 41

    Wit JM, Clayton PE, Rogol AD, Savage MO, Saenger PH, Cohen P . Idiopathic short stature: definition, epidemiology, and diagnostic evaluation. Growth Horm IGF Res 2008; 18: 89–110.

  42. 42

    Elmlinger MW, Kuhnel W, Weber MM, Ranke MB . Reference ranges for two automated chemiluminescent assays for serum insulin-like growth factor I (IGF-I) and IGF-binding protein 3 (IGFBP-3). Clin Chem Lab Med 2004; 42: 654–664.

  43. 43

    Heinze E, Holl RW . Tests of Beta Cell Function in Childhood and Adolescence. In: Ranke M, (ed). Diagnostics of Endocrine Function in Children and Adolescents. Karger: Basel, AG, 2003, pp 318–338.

  44. 44

    Dunn WB, Broadhurst D, Begley P, Zelena E, Francis-McIntyre S, Anderson N et al. Procedures for large-scale metabolic profiling of serum and plasma using gas chromatography and liquid chromatography coupled to mass spectrometry. Nat Protoc 2011; 6: 1060–1083.

  45. 45

    Whatmore AJ, Patel L, Clayton PE . A pilot study to evaluate gene expression profiles in peripheral blood mononuclear cells (PBMCs) from children with GH deficiency and Turner syndrome in response to GH treatment. Clin Endocrinol (Oxf) 2009; 70: 429–434.

  46. 46

    Stevens A, Hanson D, Whatmore A, Destenaves B, Chatelain P, Clayton P . Human growth is associated with distinct patterns of gene expression in evolutionarily conserved networks. BMC Genomics 2013; 14: 547.

  47. 47

    Stevens A, Clayton P, Tato L, Yoo HW, Rodriguez-Arnao MD, Skorodok J et al. Pharmacogenomics of insulin-like growth factor-I generation during GH treatment in children with GH deficiency or Turner syndrome. Pharmacogenomics J 2013; 14: 54–62.

  48. 48

    Sartor MA, Leikauf GD, Medvedovic M . LRpath: a logistic regression approach for identifying enriched biological groups in gene expression data. Bioinformatics 2009; 25: 211–217.

  49. 49

    Benjamini Y, Hochberg Y . Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J R Statist Soc B 1995; 57: 289–300.

  50. 50

    Calvano SE, Xiao W, Richards DR, Felciano RM, Baker HV, Cho RJ et al. A network-based analysis of systemic inflammation in humans. Nature 2005; 437: 1032–1037.

  51. 51

    Smoot ME, Ono K, Ruscheinski J, Wang PL, Ideker T . Cytoscape 2.8: new features for data integration and network visualization. Bioinformatics 2011; 27: 431–432.

  52. 52

    Deng HZ, Deng H, Su Z, Li YH, Ma HM, Chen HS et al. Insulin resistance and adiponectin levels are associated with height catch-up growth in pre-pubertal Chinese individuals born small for gestational age. Nutr Metab (Lond) 2012; 9: 107–109.

  53. 53

    Cianfarani S, Martinez C, Maiorana A, Scire G, Spadoni GL, Boemi S . Adiponectin levels are reduced in children born small for gestational age and are inversely related to postnatal catch-up growth. J Clin Endocrinol Metab 2004; 89: 1346–1351.

  54. 54

    Stefan N, Bunt JC, Salbe AD, Funahashi T, Matsuzawa Y, Tataranni PA . Plasma adiponectin concentrations in children: relationships with obesity and insulinemia. J Clin Endocrinol Metab 2002; 87: 4652–4656.

  55. 55

    Lindsay RS, Funahashi T, Hanson RL, Matsuzawa Y, Tanaka S, Tataranni PA et al. Adiponectin and development of type 2 diabetes in the Pima Indian population. Lancet 2002; 360: 57–58.

  56. 56

    Yamauchi T, Kamon J, Waki H, Terauchi Y, Kubota N, Hara K et al. The fat-derived hormone adiponectin reverses insulin resistance associated with both lipoatrophy and obesity. Nat Med 2001; 7: 941–946.

  57. 57

    Berg AH, Combs TP, Du X, Brownlee M, Scherer PE . The adipocyte-secreted protein Acrp30 enhances hepatic insulin action. Nat Med 2001; 7: 947–953.

  58. 58

    Ibanez L, Lopez-Bermejo A, Diaz M, de Zegher F . Catch-up growth in girls born small for gestational age precedes childhood progression to high adiposity. Fertil Steril 2011; 96: 220–223.

  59. 59

    Bottner A, Kratzsch J, Muller G, Kapellen TM, Blüher S, Keller E et al. Gender differences of adiponectin levels develop during the progression of puberty and are related to serum androgen levels. J Clin Endocrinol Metab 2004; 89: 4053–4061.

  60. 60

    Wishart DS, Jewison T, Guo AC, Wilson M, Knox C, Liu Y et al. HMDB 3.0—The Human Metabolome Database in 2013. Nucleic Acids Res 2013; 41 (Database issue): D801–D807.

  61. 61

    Nissen PM, Nebel C, Oksbjerg N, Bertram HC . Metabolomics reveals relationship between plasma inositols and birth weight: possible markers for fetal programming of type 2 diabetes. J Biomed Biotechnol 2011; 2011: 1147–1155.

  62. 62

    Muller G, Wied S, Crecelius A, Kessler A, Eckel J . Phosphoinositolglycan-peptides from yeast potently induce metabolic insulin actions in isolated rat adipocytes, cardiomyocytes, and diaphragms. Endocrinology 1997; 138: 3459–3475.

  63. 63

    Corrado F, D'Anna R, Di VG, Giordano D, Pintaudi B, Santamaria A et al. The effect of myoinositol supplementation on insulin resistance in patients with gestational diabetes. Diabet Med 2011; 28: 972–975.

  64. 64

    Falasca M, Hughes WE, Dominguez V, Sala G, Fostira F, Fang MQ et al. The role of phosphoinositide 3-kinase C2alpha in insulin signaling. J Biol Chem 2007; 282: 28226–28236.

  65. 65

    Islam MS, Rorsman P, Berggren PO . Ca(2+)-induced Ca2+ release in insulin-secreting cells. FEBS Lett 1992; 296: 287–291.

  66. 66

    MacDonald MJ, Smith AD III, Hasan NM, Sabat G, Fahien LA . Feasibility of pathways for transfer of acyl groups from mitochondria to the cytosol to form short chain acyl-CoAs in the pancreatic beta cell. J Biol Chem 2007; 282: 30596–30606.

  67. 67

    Kobayashi T, Kouzaki H, Kita H . Human eosinophils recognize endogenous danger signal crystalline uric acid and produce proinflammatory cytokines mediated by autocrine ATP. J Immunol 2010; 184: 6350–6358.

  68. 68

    Yuneva M, Zamboni N, Oefner P, Sachidanandam R, Lazebnik Y . Deficiency in glutamine but not glucose induces MYC-dependent apoptosis in human cells. J Cell Biol 2007; 178: 93–105.

  69. 69

    Li T, Francl JM, Boehme S, Ochoa A, Zhang Y, Klaassen CD et al. Glucose and insulin induction of bile acid synthesis: mechanisms and implication in diabetes and obesity. J Biol Chem 2012; 287: 1861–1873.

  70. 70

    McLean BN, Allen J, Ferdinandusse S, Wanders RJ . A new defect of peroxisomal function involving pristanic acid: a case report. J Neurol Neurosurg Psychiatry 2002; 72: 396–399.

  71. 71

    Mitchell CJ, Nelson AE, Cowley MJ, Kaplan W, Stone G, Sutton SK et al. Detection of growth hormone doping by gene expression profiling of peripheral blood. J Clin Endocrinol Metab 2009; 94: 4703–4709.

  72. 72

    Powell JE, Henders AK, McRae AF, Wright MJ, Martin NG, Dermitzakis ET et al. Genetic control of gene expression in whole blood and lymphoblastoid cell lines is largely independent. Genome Res 2012; 22: 456–466.

  73. 73

    Horowitz MC, Lorenzo JA . Immunologic regulation of bone development. Adv Exp Med Biol. 2007; 602: 47–56.

  74. 74

    Romanyukha AA, Yashin AI . Age related changes in population of peripheral T cells: towards a model of immunosenescence. Mech Ageing Dev 2003; 124: 433–443.

  75. 75

    Weksler ME, Szabo P . The effect of age on the B-cell repertoire. J Clin Immunol 2000; 20: 240–249.

  76. 76

    Karlberg JP, Albertsson-Wikland K, Kwan EY, Lam BC, Low LC . The timing of early postnatal catch-up growth in normal, full-term infants born short for gestational age. Horm Res 1997; 48 (Suppl 1): 17–24.

  77. 77

    Lee PA, Chernausek SD, Hokken-Koelega AC, Czernichow P . International Small for Gestational Age Advisory Board consensus development conference statement: management of short children born small for gestational age, April 24-October 1, 2001. Pediatrics 2003; 111 (6 Pt 1): 1253–1261.

Download references


AS was supported by Merck Serono SA, Geneva, Switzerland. IB was supported by an unrestricted educational grant from Novo Nordisk UK. CDL was supported by a European Society for Paediatric Endocrinology (ESPE) Research Fellowship, sponsored by Novo Nordisk A/S. WD was supported by the NIHR Manchester Biomedical Research Centre. DH was supported by a Wellcome Trust Institutional Strategic Support Fund (TSSF) award (097820) to the University of Manchester. Recruitment to this study was facilitated by the NIHR Medicines for Children Research Network.

Author information

Correspondence to P Clayton.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies the paper on the The Pharmacogenomics Journal website

Supplementary information

PowerPoint slides

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Stevens, A., Bonshek, C., Whatmore, A. et al. Insights into the pathophysiology of catch-up compared with non-catch-up growth in children born small for gestational age: an integrated analysis of metabolic and transcriptomic data. Pharmacogenomics J 14, 376–384 (2014) doi:10.1038/tpj.2014.4

Download citation


  • catch-up growth
  • metabolomics
  • network biology
  • SGA
  • transcriptomics

Further reading