You are here

Article Text

Article menu

Download PDFPDF

XML
Obstetrics and gynaecology
Research
Examining the predictive accuracy of metabolomics for small-for-gestational-age babies: a systematic review
  1. Debora Farias Batista Leite1,2,
  2. Aude-Claire Morillon3,
  3. Elias F Melo Júnior4,
  4. Renato T Souza5,
  5. Fergus P McCarthy6,
  6. Ali Khashan7,
  7. Philip Baker8,
  8. Louise C Kenny9,
  9. Jose Guilherme Cecatti10
  1. 1 Department of Tocogynecology, Campinas’ State University, Campinas, Brazil
  2. 2 Department of Maternal and Child Health, Universidade Federal de Pernambuco, Recife, Pernambuco, Brazil
  3. 3 Irish Centre for Fetal and Neonatal Translational Research (INFANT), University College Cork National University of Ireland, Cork, Ireland
  4. 4 Clinics Hospital, Universidade Federal de Pernambuco, Recife, Brazil
  5. 5 Obstetrics and Gynecology, Universidade Estadual de Campinas, Campinas, Brazil
  6. 6 Department of Gynaecology and Obstetrics, St Thomas Hospital, Cork, UK
  7. 7 Department of Epidemiology and Public Health, University College Cork, Cork, Ireland
  8. 8 College of Medicine, University of Leicester, Leicester, UK
  9. 9 Department of Women’s and Children’s Health, University of Liverpool School of Life Sciences, Liverpool, UK
  10. 10 Obstetrics and Gynecology, University of Campinas, Campinas, Sao Paolo, Brazil
  1. Correspondence to Professor Jose Guilherme Cecatti; cecatti{at}unicamp.br

Abstract

Introduction To date, there is no robust enough test to predict small-for-gestational-age (SGA) infants, who are at increased lifelong risk of morbidity and mortality.

Objective To determine the accuracy of metabolomics in predicting SGA babies and elucidate which metabolites are predictive of this condition.

Data sources Two independent researchers explored 11 electronic databases and grey literature in February 2018 and November 2018, covering publications from 1998 to 2018. Both researchers performed data extraction and quality assessment independently. A third researcher resolved discrepancies.

Study eligibility criteria Cohort or nested case–control studies were included which investigated pregnant women and performed metabolomics analysis to evaluate SGA infants. The primary outcome was birth weight <10th centile—as a surrogate for fetal growth restriction—by population-based or customised charts.

Study appraisal and synthesis methods Two independent researchers extracted data on study design, obstetric variables and sampling, metabolomics technique, chemical class of metabolites, and prediction accuracy measures. Authors were contacted to provide additional data when necessary.

Results A total of 9181 references were retrieved. Of these, 273 were duplicate, 8760 were removed by title or abstract, and 133 were excluded by full-text content. Thus, 15 studies were included. Only two studies used the fifth centile as a cut-off, and most reports sampled second-trimester pregnant women. Liquid chromatography coupled to mass spectrometry was the most common metabolomics approach. Untargeted studies in the second trimester provided the largest number of predictive metabolites, using maternal blood or hair. Fatty acids, phosphosphingolipids and amino acids were the most prevalent predictive chemical subclasses.

Conclusions and implications Significant heterogeneity of participant characteristics and methods employed among studies precluded a meta-analysis. Compounds related to lipid metabolism should be validated up to the second trimester in different settings.

PROSPERO registration number CRD42018089985.

  • small for gestational age
  • fetal growth restriction
  • metabolomics
  • prediction
  • gas-chromatography
  • mass spectrometry
  • vitamin d
  • homocysteine
  • lipids
  • fatty acids

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.

https://doi.org/10.1136/bmjopen-2019-031238

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Strengths and limitations of this study

    • To our knowledge, this is the first systematic review to assess the predictive accuracy of metabolomics for an adverse pregnancy outcome.

    • Using small for gestational age (SGA) as surrogate for fetal growth restriction—just as in epidemiological investigations—improves the translational potential of metabolomics.

    • Identification of techniques, types of maternal samples and chemical classes paves the way for future metabolomics investigations on fetal growth patterns.

    • Available data could not support a meta-analysis; further studies should include accuracy measures of individual metabolites or chemical subclasses in predicting SGA.

    Introduction

    Fetal growth restriction (FGR) and small-for-gestational-age (SGA) infants are major concerns in modern obstetrics.1–3 SGA is commonly used as a proxy for FGR,4 despite the subtle differences between these two pathological conditions. The prevalence of both varies according to the criteria applied and on the population and setting, although it reaches as much as 25% in low-income and middle-income countries.5 SGA newborns may have adverse health effects, such as stillbirth,4 perinatal asphyxia,6 impaired neurodevelopment7 and increased cardiovascular risk.8 9 To date, there are no robust prediction tools for SGA using clinical factors,10 11 ultrasound data12 13 or placental biomarkers.14

    For hypothesis-generating or validation purposes, metabolomics is a novel area of biomarker, discovery, development and clinical diagnostics in translational medicine.15 16 Metabolomics is the study of all metabolites15 16 in a given sample, that is, low molecular weight compounds (50–2000 Da) that are intermediates of biochemical reactions and metabolic pathways, considered to directly reflect cellular activity and phenotype.15 16 Recent studies have evaluated the pathophysiology17–20 of SGA with metabolomics. However, little is known about the potential of metabolomics to identify predictive compounds of SGA.

    Since metabolomics can identify multiple metabolites from low volume samples and create a model from a collection of these samples,15 it is a promising technology for hypothesis generation in a heterogeneous condition such as SGA. The prediction of SGA in pregnancy would help refer women to specialised care facilities, improving maternal and neonatal outcomes.21 22

    In this context, our review question was ‘What is the accuracy of metabolomics for predicting FGR?’. The main objective of this systematic review was to assess the accuracy of metabolomics techniques in predicting SGA. As a secondary aim, we intended to determine which metabolites are predictive of this condition.

    Methods

    The protocol for this systematic review was published previously.23 This study follows international guidelines for transparency (International Prospective Register of Systematic Reviews) and respects the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.24

    Literature search strategy

    Two independent researchers (DFBL and A-CM) assessed 11 electronic databases (PubMed, EMBASE, Latin American and Caribbean Health Sciences Literature, Scientific Electronic Library Online, Health Technology Assessment, Database of Abstracts of Reviews of Effects, Aggressive Research Intelligence Facility, Cumulative Index of Nursing and Allied Health Literature, Maternity and Infant Care, Scopus, and Web of Science) and grey literature. There were no limits or language constraints; the search strategy covered published documents between 1998 and 2018. The keywords ‘small for gestational age’, ‘metabolomics’, ‘prediction’ and ‘antenatal’, and variations of each, were combined with Boolean operators depending on each database requirements. The full EMBASE literature search was as follows: (‘fetal growth retardation’ OR ‘fetal growth restriction’ OR ‘intrauterine growth restriction’ OR ‘intrauterine growth retardation’ OR ‘small for gestational age’) AND (‘metabolomic*’ OR ‘metabonomic*’ OR ‘metabolit* ‘H NMR’ OR ‘proton NMR’ OR ‘proton nuclear magnetic resonance’ OR ‘liquid chromatogra*’ OR ‘gas chromatogra*’ OR ‘UPLC’ OR ‘ultra-performance’ OR ‘ultra performance liquid chromatograph*’) AND (‘pregnan*’ OR ‘antenat*’ OR ‘ante nat*’ OR ‘prenat*’ OR ‘pre nat*’) AND (‘screen*’ OR ‘predict*’ OR ‘metabolic profil*’). Please check online supplementary material 1 for more details.

    Supplemental material

    Outcomes and subgroup analysis

    The primary outcome was SGA, as a surrogate for FGR and defined as birth weight <10th centile, by population-based or customised charts. The secondary outcomes were birth weight ≤5th or ≤3rd centile.

    The intended subgroup analysis comprised the type of metabolomics technique applied (nuclear magnetic resonance, NMR; gas or liquid chromatography coupled with mass spectrometry, GC-MS or LC-MS, respectively); maternal health status before pregnancy (women with vs without any chronic health condition); type of SGA suspected during pregnancy (early vs late SGA); and type of pregnancy (singleton vs multiple pregnancy).

    Selection criteria of studies, data collection and analysis

    Cohort or case–control studies were included if maternal samples were collected before the clinical diagnosis of SGA, if any metabolomics technique was applied and if the results of SGA were presented. Articles presenting data from the same research project but analysing distinct metabolites or showing data from different countries were included. Studies were excluded (1) according to study design; (2) if they had not applied any metabolomics technique; (3) if they were only experimental studies; (4) if it was not possible to extract data on SGA; or (5) if they presented duplicate data, in which case the most complete publication was included for final analysis.

    Two researchers (DFBL and A-CM) independently selected studies, extracted data and discussed discrepancies. One additional reviewer (EFMJ or RTS) helped to decide, by majority, when no consensus was reached.

    Piloted standardised forms were applied for data extraction, including pregnancy characteristics and experimental details. The Human Metabolome Database (HMDB)25 and the Kyoto Encyclopedia of Genes and Genomes26 were used for matching chemical class and metabolic pathways of each metabolite, respectively.

    Risk of bias and assessment of concerns regarding applicability

    Two researchers (DFBL and A-CM) independently evaluated individual studies using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool.27 One of the third reviewers (EFMJ or RTS) helped in decision-making when no consensus was achieved.

    Each study was classified as high, low or unclear risk of bias in four domains (patient selection, index test, reference standard, and flow and timing), and as high, low or unclear concerns regarding applicability in the first three domains. We did not consider two signalling questions (‘Was a case-control design avoided?’ and ‘Was there an appropriate interval between the index test and reference standard?’). The nested case–control design was an inclusion criterion, and maternal samples should have been collected during pregnancy, that is, before the SGA diagnosis. Studies were considered ‘low risk’, for example, (1) if pregnancy or neonatal complications were not excluded in just one group of participants or data on participant selection had been provided; (2) if methods for sample preparation and interpretation were standardised or metabolite threshold was defined before the experiments (for targeted analysis); (3) if the adequacy and reasons for choosing the reference birthweight chart had been explained; or (4) if large-for-gestational-age babies had been excluded from the final comparative analysis.

    Data synthesis

    A quantitative summary of data was performed when any predictive accuracy measures could be extracted. Authors were contacted to provide additional information, when necessary. However, only Delplancke et al 28 replied. The estimation of likelihood ratios and hierarchical summary receiver operator characteristic curve29 was planned, as well as assessment of heterogeneity and publication bias.30 However, due to lack of data, a meta-analysis could not be performed.

    Patient and public involvement

    There was no patient or public involvement in conducting this systematic review.

    Results

    Literature search characteristics

    The literature search for this systematic review was performed in February 2018 and rerun in November 2018. A total of 9181 references were retrieved (figure 1). After the removal of duplicate records (n=273), title and abstract screening, and analysis of the remaining 148 full-text articles, 15 articles were included.17 18 28 31–42 See online supplementary material 2 for the excluded studies.

    Supplemental material

    Figure 1
    Figure 1

    PRISMA flow chart of study identification, screening and selection. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses. From Moher D, et al 24 For more information, visit www.prisma-statement.org.

    Characteristics of the included studies

    The characteristics of the included studies are shown in table 1. The prevalence of SGA ranged from 7.3%33 to 21.5% in cohort studies.28 There were no studies using birth weight ≤3rd centile to define SGA. The time interval between initial participant enrolment and publication varied from 317 to 54 years,40 although these data were unclear in 38% of the reports.18 28 32 33 37 In nested case–control studies, participants were matched by maternal age,17 18 38 42 ethnicity,17 18 42 parity,38 body mass index17 18 42 or infant gender.18 38

    View this table:
    Table 1

    Main characteristics of included studies

    Participant characteristics varied between studies. Regarding gestational age at assessment, samples were collected in the second trimester in half of the studies.17 18 33 35 37 39 42 In three reports, women were assessed at least twice.34 38 41 In one study, maternal blood was drawn either in the first or second trimester,40 and in another three studies only samples from the third trimester were considered.28 36 41 In the latter case, maternal hair was divided according to length, allowing evaluation of second-trimester and third-trimester metabolites.28 Studies considering the fifth centile as the cut-off sampled women in the first trimester.31 32 Twin pregnancy was a clear exclusion criterion in most studies.17 18 31 33–35 37 40–42 Pregnancy aided by assisted reproduction18 37 or women with pre-existing conditions17 18 35 37 42 were also excluded, although these data were incompletely reported.28 32 36 38 39 41 When both nulliparous and multiparous women were enrolled, there was no data analysis according to parity. Half of the studies considered term deliveries exclusively,18 28 36 38–41 and the remaining studies did not differentiate results according to gestational age at birth.

    Regarding clinical risk factors for SGA, only one paper mentioned a history of SGA, but findings were not adjusted for this variable.32 All studies, except one,28 cited participants’ smoking status. The rate of smoking habit ranged from 2.4%18 to 47.5%.40 It is important to note that Gernand et al 40 analysed samples from women recruited between 1959 and 1965, when smoking while pregnant was encouraged, which explains the high rate of smoking participants. The duration of smoking or any differences in birth weight (absolute measures or centiles) were not clearly stated. Although more prevalent in SGA pregnancies, the results did not change with this variable control.31 32 35 37 40 Only Gong et al 41 mentioned the suspicion of SGA in pregnancy, exhibiting decreasing abdominal circumference growth velocity between 20 and 36 weeks. However, on final analysis, these babies were grouped with infants not suspected during pregnancy.

    Subgroup analysis

    Due to unavailable data, the only subgroup analysis performed was related to the metabolomics approach applied (table 2). There was no mention of adherence to metabolomics reporting data guidelines. LC-MS was the leading technique used. Three studies have investigated metabolites related to environmental exposure, from contaminated water,31 consumer products36 or pesticides,42 while others have analysed endogenous compounds.32–35 37–40 Only Luthra et al 38 conducted a biomarker validation study, while Gong et al 41 chose to analyse the top 10 statistically different metabolites according to infant sex.

    View this table:
    Table 2

    Subgroup analysis of included studies according to which metabolomics technique was applied

    Maternal blood was the most common biological sample analysed by LC-MS in all studies,17 32 34–37 39–41 except for one which used GC-MS.39 Maternal urine was analysed by NMR,38 GC-MS36 or LC-MS.42 There was only one report using amniotic fluid33 and two using maternal hair,18 28 all applying GC-MS. The period of laboratory analysis was rarely specified, which made it impossible to estimate the total time of sample storage.

    Untargeted studies reported diverse metabolic features. Authors matched the peaks with an inhouse library18 28 or HMDB-related database.17 42 Horgan et al 17 found 785 compounds both in maternal and newborn samples; their predictive model included 19 metabolites (only 5 could be putatively identified; table 2) and used second-trimester maternal blood. Sulek et al 18 and Delplancke et al 28 prepared and analysed samples with GC-MS using similar protocols. Sulek et al 18 identified 32 statistically different chromatographic features from which they built a predictive model using five metabolites, including two fatty acids (2-methyloctadecanoate and margarate). In contrast, Delplancke et al 28 identified 198 metabolites, including three fatty acids (margaric, pentadecanoic and myristic acid) showing significantly higher levels in SGA cases, when second-trimester maternal hair segments were studied.

    Analysis of identified metabolites

    The identified compounds refer to 11 HMDB chemical classes. Fatty acids18 28 39 comprised the most prevalent chemical class, followed by amino acids18 33 and phosphosphingolipids17 (table 3).

    View this table:
    Table 3

    Predictive metabolites summarised according to their chemical class, subclass and biological process

    A total of 5974 women were assessed for vitamin D status. The results were presented as total vitamin D,32 35 37 40 although vitamin D2, D3 or 3-epi-25(OH)D3 35 metabolites were measured. The results were stratified according to season of maternal sampling or latitude. Either <15 ng/mL (<37.5 nmol/L)40 or <20 ng/mL (<50 nmol/L)32 35 37 levels characterised vitamin D deficiency, but were statistically different in SGA pregnancies only in the first trimester.32 Horgan et al 17 found a metabolite that could represent a vitamin D derivative, but it was only predictive in combination with 18 other compounds; this model had an area under the curve (AUC) of 0.90 (optimal OR, 44; 95% CI 9 to 214).17

    The second most frequent targeted metabolite was homocysteine,33 34 although levels were only differentiated between normal and SGA pregnancies when measured in second-trimester amniotic fluid, with a multiple linear regression model of r2=0.012 and p=0.029.33 Comparatively, the only common metabolite in the second-trimester maternal hair was margarate, with conflicting results since it was found to be either increased (AUC 0.72, 95% CI 0.58 to 0.86)28 or decreased.18 The N1,N12-diacetylspermine and the perfluorocarboxylic acids were associated with female SGA babies, not males. The former presented a fivefold decreased risk of SGA across quintiles. The perfluorodecanoic and perfluoroundecanoic acids presented OR of 3.14 (95% CI 1.07 to 9.19) and 1.83 (95% CI 1.01 to 3.32).36 Tyrosine, an essential amino acid for infants, was part of the predictive model of maternal hair, combining five metabolites with an AUC of 0.998 (95% CI 0.992 to 1.0).18 However, tyrosine did not predict SGA when urine samples were studied.38 Methylmalonic acid,34 acetate, formate or trimethylamine38 did not differentiate SGA when compared with uncomplicated pregnancies (p>0.05).

    Risk of bias and applicability concerns

    Figure 2 shows synthesised data for all included studies. See online supplementary material 3 for individual QUADAS-2 data.

    Supplemental material

    Figure 2
    Figure 2

    Assessment of risk of bias (A) and applicability concerns (B) of individual studies.

    Regarding the risk of bias, all cohort studies conducted a consecutive participant inclusion.28 33–37 39 Nested case–controls matched cases and controls randomly33–35 41 or according to maternal and infant characteristics.17 18 38 42 One study41 failed to mention matching procedures (‘Patient Selection’ domain). Researchers were not blinded to SGA status when interpreting metabolomics results,17 18 28 32 35–41 and thresholds of targeted metabolites were not prespecified31 33 36 38 39 (‘Index Test’ domain). Conversely, SGA identification was not influenced by the metabolomics test, although it was unclear when laboratory experiments were performed in some studies.18 28 31 33 34 41 Birthweight charts were adequate, except for two studies. The first did not report which centile was chosen,18 and the second used a centile designed for a different population33 (‘Reference Test’ domain). Two studies were ranked as ‘high risk’ because not all participants were included in the analysis31 37 (‘Flow and Timing’ domain).

    The QUADAS-2 tool also highlights the importance of how the findings of the included studies are suitable to the review question. In the patient selection domain, it was ranked as ‘high applicability concerns’ when infants born between the 4th and the 10th centile, but with normal abdominal circumference growth velocity, were not included in the final analysis.41 It was ‘unclear’ when the gestational age of maternal assessment was not standardised,34 or was inferred by hair segment length,28 or when few metabolites from untargeted studies were chosen for interpretation41 (‘Index Test’ domain). Finally, it was ‘high’ when the birthweight charts applied did not correspond to the study population18 33 (‘Reference Standard’ domain).

    Meta-analysis

    From the 15 included studies, only 3 were designed for prediction purposes17 18 42 and provided the AUC. The remaining reports described statistical differences of metabolites between SGA pregnancies and controls.28 31–41 Accuracy measures were extracted when available (table 2). However, due to marked heterogeneity (tables 1 and 2) of gestational age at sampling, type of samples used, type of birthweight chart chosen, thresholds for vitamin D deficiency, metabolomics approach and identified compounds, a meta-analysis could not be performed.

    Discussion

    Main findings

    In this first systematic review of metabolomics and adverse pregnancy endpoints, we presented techniques and metabolites which were studied for the prediction of SGA. Any effect on birth weight has important implications for perinatal research, since it is related to short-term and long-term outcomes,43–46 and in different generations.47 48 Intrauterine environment influences fetal growth through epigenetic processes: altered gene expression potentially leads to distinct phenotypes.49 Metabolomics is the most adequate approach to study this outcome since it is most directly related to phenotype.50

    Interpretation of metabolomics findings in pregnancy can be challenging. First, maternal metabolite concentrations are influenced by placental transfer to and from the fetus. The ‘mirror effect’, seen for maternal plasma and venous cord blood metabolites at birth,51 cannot be ruled out when only maternal specimens are studied. Second, maternal exposure to distinct compounds may affect metabolite levels. Statistically significant differences between SGA infants and controls may not express the totality of underlying pathological pathways and have no clinical meaning. Finally, it is unclear when the processes leading to SGA are initiated. The disruption in maternal metabolism can theoretically occur at any time. In general the lower the gestational age at which the condition is suspected, the more severe the phenotype will be at birth.52 53 Thus, the description of clinical data in translational studies must deal with all these confounding factors.

    Gestational age at sampling is probably the most important parameter for prediction purposes. With timely prediction, women could be referred to specialised care and have increased surveillance, and this in turn may lead to a reduction in perinatal mortality. There are temporal changes in the maternal metabolome during pregnancy28 54–57; therefore, it is reasonable to expect distinctive metabolites at different stages of pregnancy, as reported here. Unfortunately, a wide or unclear definition of gestational age of sampling34 36 38 40 renders a more precise interpretation impossible and may limit the clinical application of these results.

    In contrast, gestational age at birth and birthweight centile seem to be the hallmarks of severity and prognosis of growth restriction.6 58 Indeed, term and preterm SGA babies show distinct clinical phenotypes, and there are concerns that some babies <10th centile of birth weight are constitutionally small infants.59–61 If only term deliveries are evaluated, the most severe cases of growth restriction may be potentially missed. Moreover, when term and preterm births are analysed together, or when lower cut-offs are not specified (eg, ≤3rd or ≤5th centile), the lack of predictive metabolites might mean that they are distinct conditions. Thus, we hypothesise that the predictive performance of metabolomics may be improved if data are analysed by gestational age at delivery and by different cut-offs of birthweight centiles.

    Evidence suggests that tobacco smoke has an impact on birth weight,62–64 although it is uncertain how and when fetal growth is impaired. It is possibly related to oxidative stress,65 and both maternal and fetal metabolism may be disturbed at delivery.66 67 Studies that were included did not investigate cigarette-related chemicals or quantify exposure to tobacco smoke. Therefore, no relationship between SGA and tobacco was found. Hence, we suggest that tobacco interferes with ongoing metabolic pathological processes, or its disturbance is related to additional metabolic pathways other than the one examined by the included studies.

    Subgroup and metabolite findings

    No reports have explored data on any maternal chronic condition, suspicion of SGA in pregnancy or number of fetuses. The lack of clear statements about participant selection has hindered data interpretation and precluded these analyses.

    The majority of included studies performed a targeted approach, that is, a hypothesis-testing evaluation,16 50 driven by epidemiological or experimental data regarding SGA newborns. None of the targeted metabolites31–40 were in common with those found by ‘hypothesis-generating’ metabolic profiling17 18 28 41 42 investigations. This reinforces the suggestion that various maternal metabolic pathways may be triggered by the SGA condition and be detected by different biological samples. However, since blood is a very complex sample and GC-MS only evaluates volatile molecules,50 our findings may be biased by study methodologies.

    Untargeted studies, as expected, have characterised several metabolites that may be validated in future investigations. Nine lipids and fatty acid metabolites,17 18 28 39 two amino acids18 33 and a steroid17 32 have been identified as potential biomarkers of SGA.

    All lipid-related metabolites identified are intermediates for energy storage and breakdown. Most metabolites were found in maternal blood17 or hair of the SGA group.18 28 Blood levels of saturated and monounsaturated non-esterified fatty acids apparently remain stable throughout pregnancy, while long-chain polyunsaturated fatty acid (docosahexaenoic acid and eicosapentaenoic acid, for example) measurements seem to show ethnicity-related changes.57 Experimental data show the importance of hypoxia and oxidative stress to placental function, and ultimately to birth weight.68 69 Findings from included studies may represent a dysregulation of lipid pathways at the placental level, but an association with maternal background is unclear. Therefore, we hypothesise that disorders of lipid metabolism may be the ‘metabolic snapshot’ of defective deep placentation70 and might reflect maternal efforts to respond to impaired fetal growth.

    Recommendations on the assessment of vitamin D and cut-offs to define vitamin D deficiency in pregnancy are controversial.71 However, vitamin D supplementation decreases SGA risk.72 In early pregnancy, vitamin D status has been related to SGA,73 74 which is in accordance with this review, despite the inconsistent findings.75 There is evidence that trophoblasts actively produce and secrete vitamin D metabolites,76 but it is not clear how they mediate fetal growth impairment. Altered hepatic gene expression and liver function in vitamin D-deficient female rats77 and single nucleotide polymorphisms78 in vitamin D receptor gene have been suggested as mechanisms to be explored by a multidimensional omics approach.

    Finally, homocysteine is an intermediate metabolite of the folate cycle. It is indirectly involved with DNA methylation and is a marker of folate deficiency.79 Maternal levels rarely reach hyperhomocysteinaemia limits,80 but folate depletion81–83 and homocysteine itself80 are thought to be associated with a higher SGA risk. In this review, homocysteine was only statistically different in SGA pregnancies when measured in amniotic fluid,33 although within the normal ranges proposed for 17–21 weeks.84 Since amniocentesis is generally performed in women at higher obstetrical risk, future studies should investigate whether homocysteine in amniotic fluid represents a confounding factor or a new biomarker.85

    Methodological quality

    Most studies were ranked as ‘low risk’ of bias or applicability to the review question. However, the lack of clear descriptions of laboratory experiments, including sample preparation and storage, and blinding of the researchers to the case/control status are major pitfalls of the included studies.

    Strengths and limitations

    To our knowledge, this is the first systematic review of metabolomics and an adverse pregnancy outcome (SGA). We presented possible biomarkers of SGA pathophysiology, metabolites implicated in lipid transport and metabolic pathways, as well as gluconeogenesis.

    However, this analysis has some limitations. First, included studies showed heterogeneity, which is fundamental in systematic reviews. Indeed, there was a wide variety of participant characteristics and methods used, and not all authors provided a detailed description of methods employed. Although the Metabolomics Standards Initiative was released in 2007,86 there is still poor adherence to guidelines.87 88 Clear reporting15 87 88 and data sharing in repositories are crucial steps in identifying features of interest, specifically possible biomarkers to be validated in the clinical studies.15 Second, we could not perform a meta-analysis of the extracted data, impacting the translational potential of metabolomics.

    Third, we considered that birth weight was a surrogate measure of intrauterine development. SGA and FGR are not interchangeable concepts. However, SGA has been used as a surrogate for FGR in many clinical studies due to difficulties in defining optimal intrauterine growth: (1) FGR diagnosis relies mostly on ultrasound measurements of fetal biometry,3 89 which in turn is subject to systematic errors90; (2) intrauterine development is adaptive, rather than uniform91 or only genetically driven49; and (3) growth impairment at birth better identifies adverse neonatal outcomes than during pregnancy.58 It is recognised that changes in obstetric care occur when growth restriction is suspected, and neonatal outcomes are improved.21 22 Thus, an accurate prediction of SGA during pregnancy will be a turning point in modern obstetrics.

    Conclusions and implications for practice

    Using the available clinical tools, efforts to predict SGA remain disappointing. Since SGA is a heterogeneous condition, it benefits from metabolomics. This novel area of research allows analysis of numerous types of biological fluids and detects thousands of metabolites in complex samples.15 16 25 However, findings of this systematic review must be interpreted with caution. The type of samples used may have influenced LC-MS (second-trimester maternal blood) and GC-MS (second-trimester maternal hair) findings in individual studies. Furthermore, the prediction of SGA in the context of maternal disorders, suspected FGR and twin pregnancies is an open field for future metabolomics studies, and environmental exposure investigation as well.

    Surprisingly, none of the studies used ≤3rd centile of birth weight as a cut-off or analysed preterm deliveries and hypertensive syndromes. Considering our findings and the different phenotypic manifestations of SGA, we envision a better performance when (1) cut-offs other than the 10th centile are tested; (2) data on gestational age at sampling and at birth are standardised; and (3) other pregnancy-related syndromes are considered, especially hypertension. Thus, future metabolomics results should advance in these critical points.

    Finally, all detected biomarkers were related to lipid pathways and energy metabolism. We consider that research efforts to predict SGA should focus on compounds involved in these pathways, up to the second trimester of pregnancy.

    Acknowledgments

    We are grateful to Shauna Barret, from the Brookfield Library, University College Cork, Ireland, for her support with the literature search; Ting-Li Han, from the Department of Obstetrics, The First Affiliated Hospital of Chongqing Medical University, China, for providing additional data for this systematic review; and Luis Felipe D’Orsi, from the University of Campinas, for his support with methods’ issues.

    References

    1. 1.↵
      1. American College of Obstetricians and Gynecologists
      . ACOG practice bulletin No. 134: fetal growth restriction. Obstet Gynecol 2013;121:1122–33.doi:10.1097/01.AOG.0000429658.85846.f9
      OpenUrlCrossRefPubMed
    2. 2.↵
      2. Figueras F ,
      3. Gratacós E
      . Update on the diagnosis and classification of fetal growth restriction and proposal of a stage-based management protocol. Fetal Diagn Ther 2014;36:86–98.
      OpenUrlCrossRefPubMed
    3. 3.↵
      2. Gordijn SJ ,
      3. Beune IM ,
      4. Thilaganathan B , et al
      . Consensus definition of fetal growth restriction: a Delphi procedure. Ultrasound Obstet Gynecol 2016;48:333–9.
      OpenUrl
    4. 4.↵
      2. Bukowski R ,
      3. Hansen NI ,
      4. Willinger M , et al
      . Fetal growth and risk of stillbirth: a population-based case-control study. PLoS Med 2014;11.
    5. 5.↵
      2. ACC L ,
      3. Katz J ,
      4. Blencowe H , et al
      . National and regional estimates of term and preterm babies born small for gestational age in 138 low-income and middle-income countries in 2010. Lancet Glob Heal 2013;1:e26–36.
      OpenUrl
    6. 6.↵
      2. Mendez-Figueroa H ,
      3. Truong VTT ,
      4. Pedroza C , et al
      . Small-For-Gestational-Age infants among uncomplicated pregnancies at term: a secondary analysis of 9 Maternal-Fetal medicine units network studies. Am J Obstet Gynecol 2016;215:628.e1–628.e7.
      OpenUrl
    7. 7.↵
      2. Sharma D ,
      3. Farahbakhsh N ,
      4. Shastri S , et al
      . Intrauterine growth restriction–part 2. J Matern Neonatal Med 2016;29:4037–48.
      OpenUrl
    8. 8.↵
      2. Barker DP ,
      3. Osmond C ,
      4. Simmonds SJ , et al
      . The relation of small head circumference and thinness at birth to death from cardiovascular disease in adult life. Br Med J 1993;306:422–6.
      OpenUrlAbstract/FREE Full Text
    9. 9.↵
      2. Balasuriya CND ,
      3. Stunes AK ,
      4. Mosti MP , et al
      . Metabolic outcomes in adults born preterm with very low birth weight or small for gestational age at term: a cohort study. J Clin Endocrinol Metab 2018;103:4437–46.
      OpenUrl
    10. 10.↵
      2. Goto E
      . Maternal anthropometry to predict small for gestational age: a meta-analysis. Eur J Obstet Gynecol Reprod Biol 2016;203:193–8.
      OpenUrl
    11. 11.↵
      2. ASD P ,
      3. Frøen JF ,
      4. Staff AC , et al
      . Prediction of small-for-gestational-age status by symphysis–fundus height: a registry-based population cohort study. BJOG 2016;123:1167–73.
      OpenUrl
    12. 12.↵
      2. Parry S ,
      3. Sciscione A ,
      4. Haas DM , et al
      . Role of early second-trimester uterine artery Doppler screening to predict small-for-gestational-age babies in nulliparous women. Am J Obstet Gynecol 2017;217:594.e1–594.e10.
      OpenUrl
    13. 13.↵
      2. Roma E ,
      3. Arnau A ,
      4. Berdala R , et al
      . Ultrasound screening for fetal growth restriction at 36 vs 32 weeks’ gestation: A randomized trial (ROUTE). Ultrasound Obstet Gynecol 2015;46:391–7.
      OpenUrl
    14. 14.↵
      2. Conde-Agudelo A ,
      3. Papageorghiou a T ,
      4. Kennedy SH , et al
      . Novel biomarkers for predicting intrauterine growth restriction: a systematic review and meta-analysis. BJOG 2013;120:681–94.
      OpenUrlCrossRefPubMed
    15. 15.↵
      2. Xia J ,
      3. Broadhurst DI ,
      4. Wilson M , et al
      . Translational biomarker discovery in clinical metabolomics: an introductory tutorial. Metabolomics 2013;9:280–99.
      OpenUrlCrossRefPubMedWeb of Science
    16. 16.↵
      2. Patti GJ ,
      3. Yanes O ,
      4. Siuzdak G
      . Metabolomics: the apogee of the omics trilogy. Nat Rev Mol cell Biol 2012;13:263–9.
      OpenUrlCrossRefPubMed
    17. 17.↵
      2. Horgan RP ,
      3. Broadhurst DI ,
      4. Walsh SK , et al
      . Metabolic profiling uncovers a phenotypic signature of small for gestational age in early pregnancy. J Proteome Res 2011;10:3660–73.
      OpenUrlCrossRefPubMed
    18. 18.↵
      2. Sulek K ,
      3. Han T-L ,
      4. Villas-Boas SG , et al
      . Hair metabolomics: identification of fetal compromise provides proof of concept for biomarker discovery. Theranostics 2014;4:953–9.doi:10.7150/thno.9265
      OpenUrlCrossRefPubMed
    19. 19.↵
      2. Favretto D ,
      3. Cosmi E ,
      4. Ragazzi E , et al
      . Cord blood metabolomic profiling in intrauterine growth restriction. Anal Bioanal Chem 2012;402:1109–21.
      OpenUrlCrossRefPubMed
    20. 20.↵
      2. Sanz-Cortés M ,
      3. Carbajo RJ ,
      4. Crispi F , et al
      . Metabolomic profile of umbilical cord blood plasma from early and late intrauterine growth restricted (IUGR) neonates with and without signs of brain vasodilation. PLoS One 2013;8:e80121.
    21. 21.↵
      2. Monier I ,
      3. Blondel B ,
      4. Ego A , et al
      . Does the presence of risk factors for fetal growth restriction increase the probability of antenatal detection? A French national study. Paediatr Perinat Epidemiol 2016;30:46–55.
      OpenUrl
    22. 22.↵
      2. Verlijsdonk JW ,
      3. Winkens B ,
      4. Boers K , et al
      . Suspected versus non-suspected small-for-gestational age fetuses at term: perinatal outcomes. J Matern Neonatal Med 2012;25:938–43.
      OpenUrl
    23. 23.↵
      2. Leite DFB ,
      3. Morillon A-C ,
      4. Melo Júnior EF , et al
      . Metabolomics for predicting fetal growth restriction: protocol for a systematic review and meta-analysis. BMJ Open 2018;8:e022743.
    24. 24.↵
      2. Moher D ,
      3. Liberati A ,
      4. Tetzlaff J , et al
      . Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med 2009;151:264–9.
      OpenUrlCrossRefPubMedWeb of Science
    25. 25.↵
      2. Wishart DS ,
      3. Feunang YD ,
      4. Marcu A , et al
      . HMDB 4.0: the human metabolome database for 2018. Nucleic Acids Res 2018;46:D608–17.
      OpenUrlCrossRefPubMed
    26. 26.↵
      1. KEGG
      . Kyoto encyclopedia of genes and genomes. Available: https://www.genome.jp/kegg/ [Accessed 20 Dec 2018].
    27. 27.↵
      2. Whiting PF ,
      3. Rutjes AWS ,
      4. Westwood ME , et al
      . Research and reporting methods accuracy studies. Ann Intern Med 2011;155:529–36.
      OpenUrlCrossRefPubMedWeb of Science
    28. 28.↵
      2. Delplancke TDJ ,
      3. De Seymour J ,
      4. Tong C , et al
      . Analysis of sequential hair segments reflects changes in the metabolome across the trimesters of pregnancy. Sci Rep 2018;8:1–12.
      OpenUrlCrossRefPubMed
    29. 29.↵
      2. Rutter CM ,
      3. Gatsonis CA
      . A hierarchical regression approach to meta-analysis of diagnostic test accuracy evaluations. Stat Med 2001;20:2865–84.
      OpenUrlCrossRefPubMedWeb of Science
    30. 30.↵
      2. Deeks JJ ,
      3. Macaskill P ,
      4. Irwig L
      . The performance of tests of publication bias and other sample size effects in systematic reviews of diagnostic test accuracy was assessed. J Clin Epidemiol 2005;58:882–93.
      OpenUrlCrossRefPubMedWeb of Science
    31. 31.↵
      2. Costet N ,
      3. Garlantézec R ,
      4. Monfort C , et al
      . Environmental and urinary markers of prenatal exposure to drinking water disinfection by-products, fetal growth, and duration of gestation in the PELAGIE birth cohort (Brittany, France, 2002-2006). Am J Epidemiol 2012;175:263–75.
      OpenUrlCrossRefPubMedWeb of Science
    32. 32.↵
      2. Ertl R ,
      3. CKH Y ,
      4. Samaha R , et al
      . Maternal serum vitamin D at 11-13 weeks in pregnancies delivering small for gestational age neonates. Fetal Diagn Ther 2012;31:103–8.
      OpenUrlCrossRefPubMed
    33. 33.↵
      2. Grandone E ,
      3. Colaizzo D ,
      4. Vecchione G , et al
      . Homocysteine levels in amniotic fluid. Thromb Haemost 2006;95:625–8.
      OpenUrlPubMed
    34. 34.↵
      2. Choi R ,
      3. Choi S ,
      4. Lim Y , et al
      . A prospective study on serum methylmalonic acid and homocysteine in pregnant women. Nutrients 2016;8:1–15.
      OpenUrlCrossRefPubMed
    35. 35.↵
      2. Kiely ME ,
      3. Zhang JY ,
      4. Kinsella M , et al
      . Vitamin D status is associated with uteroplacental dysfunction indicated by pre-eclampsia and small-for-gestational-age birth in a large prospective pregnancy cohort in Ireland with low vitamin D status. Am J Clin Nutr 2016;104:354–61.
      OpenUrlAbstract/FREE Full Text
    36. 36.↵
      2. Wang Y ,
      3. Adgent M ,
      4. PH S , et al
      . Prenatal exposure to perfluorocarboxylic acids (PFCAs) and fetal and postnatal growth in the Taiwan maternal and infant cohort study. Environ Health Perspect 2016;124:1794–800.
      OpenUrl
    37. 37.↵
      2. Ong YL ,
      3. Quah PL ,
      4. Tint MT , et al
      . The association of maternal vitamin D status with infant birth outcomes, postnatal growth and adiposity in the first 2 years of life in a multi-ethnic Asian population: the growing up in Singapore towards healthy outcomes (GUSTO) cohort study. Br J Nutr 2016;116:621–31.
      OpenUrl
    38. 38.↵
      2. Luthra G ,
      3. Vuckovic I ,
      4. Bangdiwala A , et al
      . First and second trimester urinary metabolic profiles and fetal growth restriction: an exploratory nested case-control study within the infant development and environment study. BMC Pregnancy Childbirth 2018;18:1–8.
      OpenUrl
    39. 39.↵
      2. van Eijsden M ,
      3. Hornstra G ,
      4. van der Wal MF , et al
      . Maternal n-3, n-6, and trans fatty acid profile early in pregnancy and term birth weight: a prospective cohort study. Am J Clin Nutr 2008;87:887–95.
      OpenUrlAbstract/FREE Full Text
    40. 40.↵
      2. Gernand AD ,
      3. Simhan HN ,
      4. Klebanoff MA , et al
      . Maternal serum 25-hydroxyvitamin D and measures of newborn and placental weight in a U.S. multicenter cohort study. J Clin Endocrinol Metab 2013;98:398–404.
      OpenUrlCrossRefPubMedWeb of Science
    41. 41.↵
      2. Gong S ,
      3. Sovio U ,
      4. Aye I , et al
      . Placental polyamine metabolism differs by fetal sex, fetal growth restriction, and preeclampsia. JCI Insight 2018;3:1–15.
      OpenUrl
    42. 42.↵
      2. Morillon A-C ,
      3. Yakkundi S ,
      4. Thomas G , et al
      . Untargeted UPLC-MS analysis of potential pesticide and biomarkers of fetal growth restriction. Conf Proceedings 14th Annu Conf Metabolomics Soc, 2018:200.
    43. 43.↵
      2. Wang N ,
      3. Wang X ,
      4. Li Q , et al
      . The famine exposure in early life and metabolic syndrome in adulthood. Clin Nutr 2017;36:253–9.
      OpenUrl
    44. 44.↵
      2. Hales CN ,
      3. Barker DJP
      . The thrifty phenotype hypothesis. Br Med Bull 2001;60:5–20.
      OpenUrlCrossRefPubMedWeb of Science
    45. 45.↵
      2. Melo AS ,
      3. Vieira CS ,
      4. Barbieri MA , et al
      . High prevalence of polycystic ovary syndrome in women born small for gestational age. Hum Reprod 2010;25:2124–31.
      OpenUrlCrossRefPubMedWeb of Science
    46. 46.↵
      2. Ravelli AC ,
      3. van der Meulen JH ,
      4. Michels RP , et al
      . Glucose tolerance in adults after prenatal exposure to famine. Lancet 1998;351:173–7.doi:10.1016/s0140-6736(97)07244-9
      OpenUrlCrossRefPubMedWeb of Science
    47. 47.↵
      2. Chamorro-Garcia R ,
      3. Diaz-Castillo C ,
      4. Shoucri BM , et al
      . Ancestral perinatal obesogen exposure results in a transgenerational thrifty phenotype in mice. Nat Commun 2017;8.
    48. 48.↵
      2. Seferovic MD ,
      3. Goodspeed DM ,
      4. Chu DM , et al
      . Heritable IUGR and adult metabolic syndrome are reversible and associated with alterations in the metabolome following dietary supplementation of 1-carbon intermediates. Faseb J 2015;29:2640–52.
      OpenUrlCrossRefPubMed
    49. 49.↵
      2. Padmanabhan V ,
      3. Cardoso RC ,
      4. Puttabyatappa M , et al
      . A pathway to disease. Endocrinology 2016;157:1328–40.
      OpenUrlCrossRefPubMed
    50. 50.↵
      2. Dunn WB ,
      3. Broadhurst DI ,
      4. Atherton HJ , et al
      . Systems level studies of mammalian metabolomes: the roles of mass spectrometry and nuclear magnetic resonance spectroscopy. Chem Soc Rev 2011;40:387–426.
      OpenUrlCrossRefPubMedWeb of Science
    51. 51.↵
      2. Visentin S ,
      3. Crotti S ,
      4. Donazzolo E , et al
      . Medium chain fatty acids in intrauterine growth restricted and small for gestational age pregnancies. Metabolomics 2017;13:1–9.
      OpenUrlCrossRef
    52. 52.↵
      2. Korzeniewski SJ ,
      3. Allred EN ,
      4. Joseph RM , et al
      . Neurodevelopment at age 10 years of children born. Pediatrics 2017;140.
    53. 53.↵
      2. Lees C ,
      3. Marlow N ,
      4. Arabin B , et al
      . Perinatal morbidity and mortality in early-onset fetal growth restriction: cohort outcomes of the trial of randomized umbilical and fetal flow in Europe (TRUFFLE). Ultrasound Obstet Gynecol 2013;42:400–8.
      OpenUrlCrossRefPubMedWeb of Science
    54. 54.↵
      2. Luan H ,
      3. Meng N ,
      4. Liu P , et al
      . Non-Targeted metabolomics and lipidomics LC-MS data from maternal plasma of 180 healthy pregnant women. Gigascience 2015;4:16–19.
      OpenUrl
    55. 55.↵
      2. Di Giulio AM ,
      3. Carelli S ,
      4. Castoldi RE , et al
      . Plasma amino acid concentrations throughout normal pregnancy and early stages of intrauterine growth restricted pregnancy. J Matern Neonatal Med 2004;15:356–62.
      OpenUrl
    56. 56.↵
      2. Orczyk-Pawilowicz M ,
      3. Jawien E ,
      4. Deja S , et al
      . Metabolomics of human amniotic fluid and maternal plasma during normal pregnancy. PLoS One 2016;11:1–13.
      OpenUrlCrossRefPubMed
    57. 57.↵
      2. Lindsay KL ,
      3. Hellmuth C ,
      4. Uhl O , et al
      . Longitudinal metabolomic profiling of amino acids and lipids across healthy pregnancy. PLoS One 2015;10:1–19.
      OpenUrlCrossRefPubMed
    58. 58.↵
      2. Boghossian NS ,
      3. Geraci M ,
      4. Edwards EM , et al
      . Neonatal and fetal growth charts to identify preterm infants. Am J Obstet Gynecol 2018;219:195.e1–195.e14.
      OpenUrl
    59. 59.↵
      2. Katz J ,
      3. LA W ,
      4. Mullany LC , et al
      . Prevalence of small-for-gestational-age and its mortality risk varies by choice of birth-weight-for-gestation reference population. PLoS One 2014;9:1–9.
      OpenUrlCrossRefPubMed
    60. 60.↵
      2. Ray JG ,
      3. Jiang D ,
      4. Sgro M , et al
      . Thresholds for small for gestational age among newborns of East Asian and South Asian ancestry. J Obstet Gynaecol Canada 2009;31:322–30.
      OpenUrl
    61. 61.↵
      2. Cheng YKY ,
      3. Leung TY ,
      4. TTH L , et al
      . Impact of replacing Chinese ethnicity-specific fetal biometry charts with the INTERGROWTH-21ststandard. BJOG An Int J Obstet Gynaecol 2016;123:48–55.
      OpenUrl
    62. 62.↵
      2. Abraham M ,
      3. Alramadhan S ,
      4. Iniguez C , et al
      . A systematic review of maternal smoking during pregnancy and fetal measurements with meta-analysis. PLoS One 2017;12:1–13.
      OpenUrlCrossRefPubMed
    63. 63.↵
      2. Vesterinen HM ,
      3. Morello-Frosch R ,
      4. Sen S , et al
      . Cumulative effects of prenatal-exposure to exogenous chemicals and psychosocial stress on fetal growth: Systematic-review of the human and animal evidence. PLoS One 2017;12.
    64. 64.↵
      2. Dessì A ,
      3. Corona L ,
      4. Pintus R , et al
      . Exposure to tobacco smoke and low birth weight: from epidemiology to metabolomics. Expert Rev Proteomics 2018;15:647–56.
      OpenUrl
    65. 65.↵
      2. Stone WL ,
      3. Bailey B ,
      4. Khaisha N
      . The pathophysiology of smoking during pregnancy: a systems biology approach. Front Biosci 2014;E6:318–28.
      OpenUrl
    66. 66.↵
      2. Fischera T ,
      3. Lilic LN ,
      4. Lic S , et al
      . Low-Level maternal exposure to nicotine associates with significant metabolic perturbations in second-trimester amniotic fluid. Env Int 2017;107:227–34.
      OpenUrl
    67. 67.↵
      2. Rolle-Kampczyk UE ,
      3. Krumsiek J ,
      4. Otto W , et al
      . Metabolomics reveals effects of maternal smoking on endogenous metabolites from lipid metabolism in cord blood of newborns. Metabolomics 2016;12.
    68. 68.↵
      2. Thomas MM ,
      3. Haghiac M ,
      4. Grozav C , et al
      . Oxidative stress impairs fatty acid oxidation and mitochondrial function in the term placenta. Reprod Sci 2018;193371911880205.
    69. 69.↵
      2. Määttä J ,
      3. Sissala N ,
      4. Dimova EY , et al
      . Hypoxia causes reductions in birth weight by altering maternal glucose and lipid metabolism. Sci Rep 2018;8.
    70. 70.↵
      2. Brosens I ,
      3. Pijnenborg R ,
      4. Vercruysse L , et al
      . The ‘great obstetrical syndromes’ are associated with disorders of deep placentation. Am J Obs Gynecol 2011;204:193–201.
      OpenUrlCrossRefPubMedWeb of Science
    71. 71.↵
      2. Holick MF ,
      3. Binkley NC ,
      4. Bischoff-Ferrari HA , et al
      . Evaluation, treatment, and prevention of vitamin D deficiency: an endocrine Society clinical practice guideline. J Clin Endocrinol Metab 2011;96:1911–30.
      OpenUrlCrossRefPubMedWeb of Science
    72. 72.↵
      2. WG B ,
      3. Nuyt AM ,
      4. Weiler H , et al
      . Association between vitamin D supplementation during pregnancy and offspring growth, morbidity, and mortality: a systematic review and meta-analysis. JAMA Pediatr 2018;172:635–45.
      OpenUrl
    73. 73.↵
      2. Bodnar LM ,
      3. Catov JM ,
      4. Zmuda JM , et al
      . Maternal serum 25-hydroxyvitamin D concentrations are associated with Small-for-Gestational age births in white women. J Nutr 2010;140:999–1006.
      OpenUrlAbstract/FREE Full Text
    74. 74.↵
      2. Leffelaar ER ,
      3. Vrijkotte TGM ,
      4. Van Eijsden M
      . Maternal early pregnancy vitamin D status in relation to fetal and neonatal growth: results of the multi-ethnic Amsterdam born children and their development cohort. Br J Nutr 2010;104:108–17.
      OpenUrlCrossRefPubMed
    75. 75.↵
      2. Martínez-Domínguez SJ ,
      3. Tajada M ,
      4. Chedraui P , et al
      . Systematic review and meta-analysis of Spanish studies regarding the association between maternal 25-hydroxyvitamin D levels and perinatal outcomes. Gynecol Endocrinol 2018:1–8.
    76. 76.↵
      2. Park H ,
      3. Wood MR ,
      4. Malysheva O , et al
      . Placental vitamin D metabolism and its associations with circulating vitamin D metabolites in pregnant women. Am J Clin Nutr 2017;106:1439–48.
      OpenUrlAbstract/FREE Full Text
    77. 77.↵
      2. Sharma SS ,
      3. Jangale NM ,
      4. Harsulkar AM , et al
      . Chronic maternal calcium and 25-hydroxyvitamin D deficiency in Wistar rats programs abnormal hepatic gene expression leading to hepatic steatosis in female offspring, 2017.
    78. 78.↵
      2. Barchitta M ,
      3. Maugeri A ,
      4. La Rosa MC , et al
      . Single nucleotide polymorphisms in vitamin D receptor gene affect birth weight and the risk of preterm birth: Results from the “mamma & bambino” cohort and a meta-analysis. Nutrients 2018;10:18.
    79. 79.↵
      2. Kim J ,
      3. Kim H ,
      4. Roh H , et al
      . Causes of hyperhomocysteinemia and its pathological significance. Arch Pharm Res 2018;41:372–83.
      OpenUrl
    80. 80.↵
      2. Yajnik CS ,
      3. Chandak GR ,
      4. Joglekar C , et al
      . Maternal homocysteine in pregnancy and offspring birthweight: epidemiological associations and Mendelian randomization analysis. Int J Epidemiol 2014;43:1487–97.
      OpenUrlCrossRefPubMed
    81. 81.↵
      2. Smits LJM ,
      3. Essed GGM
      . Short interpregnancy intervals and unfavourable pregnancy outcome: role of folate depletion. Lancet 2001;358:2074–7.
      OpenUrlCrossRefPubMedWeb of Science
    82. 82.↵
      2. van Eijsden M ,
      3. van der Wal MF ,
      4. Bonsel GJ
      . Association between short interpregnancy intervals and term birth. Am J Clin Nutr 2008;88:147–53.
      OpenUrlAbstract/FREE Full Text
    83. 83.↵
      2. Hogeveen M ,
      3. Blom HJ ,
      4. Den Heijer M
      . Maternal homocysteine and small-for-gestational-age offspring: systematic review and meta-analysis. Am J Clin Nutr 2012;95:130–6.
      OpenUrlAbstract/FREE Full Text
    84. 84.↵
      2. Imbard A ,
      3. Blom HJ ,
      4. Schlemmer D , et al
      . Methylation metabolites in amniotic fluid depend on gestational age. Prenat Diagn 2013;33:848–55.
      OpenUrlCrossRefPubMed
    85. 85.↵
      2. Heazell A ,
      3. Newman L ,
      4. Lean S , et al
      . Pregnancy outcome in mothers over the age of 35. Curr Opin Obs Gynecol 2018;30:337–43.
      OpenUrl
    86. 86.↵
      2. Goodacre R ,
      3. Broadhurst D ,
      4. Smilde AK , et al
      . Proposed minimum reporting standards for data analysis in metabolomics. Metabolomics 2007;3:231–41.
      OpenUrlCrossRefWeb of Science
    87. 87.↵
      2. Considine EC ,
      3. Thomas G ,
      4. Boulesteix AL , et al
      . Critical review of reporting of the data analysis step in metabolomics. Metabolomics 2018;14.
    88. 88.↵
      2. Spicer RA ,
      3. Salek R ,
      4. Steinbeck C
      . Compliance with minimum information guidelines in public metabolomics repositories. Sci Data 2017;4:1–8.
      OpenUrlCrossRef
    89. 89.↵
      2. McCowan LM ,
      3. Figueras F ,
      4. Anderson NH
      . Evidence-Based national guidelines for the management of suspected fetal growth restriction: comparison, consensus, and controversy. Am J Obstet Gynecol 2018;218:S855–68.
      OpenUrl
    90. 90.↵
      2. Lappen JR ,
      3. Myers SA
      . The systematic error in the estimation of fetal weight and the underestimation of fetal growth restriction. Am J Obstet Gynecol 2017;216:477–83.
      OpenUrl
    91. 91.↵
      2. Papageorghiou AT ,
      3. Ohuma EO ,
      4. Altman DG , et al
      . International standards for fetal growth based on serial ultrasound measurements: the fetal growth longitudinal study of the Intergrowth-21st project. Lancet 2014;384:869–79.
      OpenUrlCrossRefPubMed

    Footnotes

    • Contributors DFBL and A-CM have equally contributed to this report, and both are guarantors of this review. They elaborated on the protocol, searched the literature, selected studies, extracted data, assessed risk of bias and drafted the initial manuscript. RTS and EFMJ have participated in judging inclusion of studies, interpreting data and revising the manuscript. FPM has supported data extraction and has critically examined the clinical interpretation of the results. AK has discussed the quantitative data synthesis and supervised the report writing. PB, LCK and JGC have supervised and approved all steps. All authors have read and agree with this submission.

    • Funding DFBL (process number 88881.134512/2016-01) and RTS (88881.134095/2016-01) have scholarships awarded by Brazilian Federal Agency for Support and Evaluation of Graduate Education (CAPES). A-CM was granted a scholarship from Science Foundation Ireland for her doctoral thesis. PRETERM-SAMBA has granted sponsor from Brazilian National Research Council (CNPq) (Award 401636/2013-5) and from the Bill and Melinda Gates Foundation (grant OPP1107597), corresponding to the research call 'Grand Challenges Brazil: Reducing the burden of preterm birth', number 05/2013. This research received no specific grant from commercial or not-for-profit sectors. The sponsors have not intervened in the authors’ decision to write the systematic review protocol or to submit this paper.

    • Competing interests None declared.

    • Patient consent for publication Not required.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data availability statement No data are available.