Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Is there a maternal blood biomarker that can predict spontaneous preterm birth prior to labour onset? A systematic review

  • Kylie K. Hornaday,

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Department of Physiology and Pharmacology, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada

    ⨯
  • Eilidh M. Wood,

    Roles Data curation, Writing – review & editing

    Affiliation Department of Physiology and Pharmacology, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada

    ⨯
  • Donna M. Slater

    Roles Conceptualization, Investigation, Methodology, Supervision, Writing – review & editing

    dmslater@ucalgary.ca

    Affiliations Department of Physiology and Pharmacology, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada, Department of Obstetrics and Gynecology, Cumming School of Medicine, University of Calgary, Calgary, Alberta, Canada

    ⨯

Abstract

Introduction

The ability to predict spontaneous preterm birth (sPTB) prior to labour onset is a challenge, and it is currently unclear which biomarker(s), may be potentially predictive of sPTB, and whether their predictive power has any utility. A systematic review was conducted to identify maternal blood biomarkers of sPTB.

Methods

This study was conducted according to PRISMA protocol for systematic reviews. Four databases (MEDLINE, EMBASE, CINAHL, Scopus) were searched up to September 2021 using search terms: “preterm labor”, “biomarker” and “blood OR serum OR plasma”. Studies assessing blood biomarkers prior to labour onset against the outcome sPTB were eligible for inclusion. Risk of bias was assessed based on the Newcastle Ottawa scale. Increased odds of sPTB associated with maternal blood biomarkers, as reported by odds ratios (OR), or predictive scores were synthesized. This review was not prospectively registered.

Results

Seventy-seven primary research articles met the inclusion criteria, reporting 278 unique markers significantly associated with and/or predictive of sPTB in at least one study. The most frequently investigated biomarkers were those measured during maternal serum screen tests for aneuploidy, or inflammatory cytokines, though no single biomarker was clearly predictive of sPTB based on the synthesized evidence. Immune and signaling pathways were enriched within the set of biomarkers and both at the level of protein and gene expression.

Conclusion

There is currently no known predictive biomarker for sPTB. Inflammatory and immune biomarkers show promise, but positive reporting bias limits the utility of results. The biomarkers identified may be more predictive in multi-marker models instead of as single predictors. Omics-style studies provide promising avenues for the identification of novel (and multiple) biomarkers. This will require larger studies with adequate power, with consideration of gestational age and the heterogeneity of sPTB to identify a set of biomarkers predictive of sPTB.

Citation: Hornaday KK, Wood EM, Slater DM (2022) Is there a maternal blood biomarker that can predict spontaneous preterm birth prior to labour onset? A systematic review. PLoS ONE 17(4): e0265853. https://doi.org/10.1371/journal.pone.0265853

Editor: Clive J. Petry, University of Cambridge, UNITED KINGDOM

Received: November 10, 2021; Accepted: March 8, 2022; Published: April 4, 2022

Copyright: © 2022 Hornaday et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript and its Supporting information files.

Funding: KKH is funded by the University of Calgary Eyes High Doctoral Recruitment Scholarship EMW and DMS are funded by the Canadian Institute of Health Research The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Globally, over 15 million babies are born prematurely each year, a value that, according to the Global Disease Burden study (2010–2019) has remained relatively consistent over the last decade. While neonatal mortality rates have dropped, largely attributed to advances in neonatal care, preterm birth continues to be associated with lifelong morbidities and is estimated to account for over 69 million disability-adjusted life years (DALYs) annually worldwide (Global Disease Burden Study, 2019). An estimated 45% of all preterm births occur following spontaneous onset of labour with no known maternal or fetal indications [1]. The etiology of a spontaneous preterm birth (sPTB) remains unclear, limiting strategies for prediction and management. Early prediction would allow for targeted clinical management, such as allowing time for the administration of progesterone or cervical cerclage and provide valuable reassurance to those identified as low risk for premature delivery.

Early prediction strategies are particularly limited for asymptomatic populations. Cervicovaginal levels of fetal fibronectin (fFN test) is used to predict imminent delivery (within seven days) [2]. The fFN test is commonly ordered for those 22–35 weeks pregnant with signs and symptoms of labour but is not generally recommended when asymptomatic for labour [3]. The strongest predictor of sPTB currently available is a previous preterm birth, which is associated with a 23% chance of another preterm birth [4]. While this can help guide clinical management in a subsequent pregnancy, this cannot help those at risk for a first-time sPTB. There is increasing evidence that biomarkers, or any biological entity or characteristic such as proteins or metabolites, in maternal circulation during pregnancy are associated with subsequent onset of sPTB [5, 6]. Biomarkers within amniotic fluid have been investigated for early prediction of sPTB [7], but amniocentesis can introduce risk of intrauterine infection and other unwanted complications [8]. There is an ongoing need for both more accessible and reliable tests for early prediction of sPTB, which would identify those at-risk and who may benefit from early intervention.

Peripheral blood is routinely collected during antenatal care and thus presents an alluring opportunity for a relatively non-invasive tool for prediction. This has garnered extensive studies on whether a blood test can accurately assess risk of subsequent preterm birth, including increasingly accessible high-throughput and -‘omics’ techniques, which have provided unprecedented amounts of data that can be used to predict health outcomes [9]. The aim of this review was to aggregate and synthesize existing data on biomarkers for spontaneous preterm birth, including to identify and synthesize biomarkers associated with or predictive of spontaneous preterm birth.

Methods

Research question

What biomarkers are predictive of, or associated with sPTB when there are no prior symptoms of labour?

Review method

The search strategy, study selection, and data extraction protocols undertaken in this study were developed according to the PRISMA guidelines for systematic reviews [10]. The protocol for this study was not prospectively registered.

Eligibility criteria

All observational studies, for example, primary case-control, cross-sectional or cohort studies were included. Literature reviews, systematic reviews with no meta-analysis, or incomplete publication (e.g. conference abstract) were excluded, as well as non-human studies. Eligible studies are those which investigated the association of maternal blood biomarkers for sPTB or developed a prediction model for sPTB using maternal blood biomarkers. Maternal samples included whole blood, plasma, or serum. Eligible biomarkers included biochemical and molecular biomarkers including, but not limited to, proteins, nucleic acids, or metabolites. Genetic and epi-genetic variant (single nucleotide polymorphisms or other) markers were excluded. At the time of sample collection, study participants must have no signs of labour, including but not limited to uterine contractions and rupture of membranes. Studies investigating risk of sPTB in multiple pregnancies, those that did not have a clear and standardized definition of sPTB as the outcome of interest, and provider initiated preterm births, otherwise known as medically indicated preterm births were also excluded.

Search strategy

Four databases were searched for records: MEDLINE, EMBASE, CINAHL and Scopus. Additional studies were collected by searching reference lists of records and relevant review articles and using manual search of Google Scholar. The initial search was conducted on March 21, 2019, then repeated September 13th, 2021, to identify new publications. There was no time restriction to the search, though time frames vary by database (earliest date range 1788 Scopus– 1974 EMBASE). Only studies published in English, or which have an available English translation were considered. The search strategy included the following terms “spontaneous preterm birth” AND “biomarker” AND “blood”, including all relevant synonyms and alternate terms. Search terms and syntax were adjusted accordingly for each database, including the addition of relevant Medical Subject Headings (MeSH) where appropriate, and the searches were kept as similar as possible. No additional filters were applied to the searches. Complete search input for each database is described in S1 File.

Study selection

Records retrieved were exported to a citation manager (Endnote Web). Duplicates were removed and the remaining articles were initially screened by title and abstract, those not relevant to the research question were removed. Remaining full-text articles were then assessed against the eligibility criteria. The reason for exclusion was noted for ineligible studies. Two investigators independently assessed the articles identified using the same search terms (KKH and EMW), and when a consensus could not be reached, a third reviewer acted as a tiebreaker (DMS).

Quality assessment

The Newcastle-Ottawa Scale (NOS) scoring system was used to assess methodological quality and risk of bias for all eligible studies (S2 File). Case-control studies were assessed based on 1) adequate case definition for sPTB (in the case of multiple case groups, the paper was scored on only the case definition for sPTB), 2) representativeness of the cases, 3) selection of controls, 4) definition of controls with respect to history of sPTB, 5) comparability of cases and controls, 6) ascertainment of exposure/biomarker and 7) whether the same method of ascertainment for biomarker measurement was used in both cases and controls. Cohort and cross-sectional studies were assessed based on 1) cohort representativeness of the pregnant community, 2) assessment of outcome (in the case of multiple outcomes, the paper was scored on only the outcome definition for sPTB) and 3) adequacy of follow up. Methodological quality was independently rated by two reviewers, KKH and EMW. In the case of disagreement, ratings were discussed, and a consensus reached. Studies with total NOS scores below 50% were excluded from subsequent data extraction and synthesis.

Data extraction

Data was extracted from eligible studies using a standardized template adapted from the Joanna Briggs Institute (JBI) data extraction form. The following data was collected from each study by a single reviewer (KH): objectives, participant characteristics, participant numbers, study setting, study year range, tissue (blood, serum or plasma), method of measurement of biomarker, timepoint of measurement, outcome of interest, country of origin, statistical test, effect measures as reported by odds ratios (OR), predictive value as reported by area under the receiver operator curve (AUC), biomarker levels in sPTB and term populations, results/direction, and comments pertaining to the heterogeneity of results. Missing or unclear information was marked ‘not stated’.

Enrichment analysis of biomarkers

Top biomarkers were analyzed to identify common pathways or processes of interest. The goal of pathway analysis is to detect relevant groups of genes or proteins that are commonly associated with a biological function or process [11]. Genes or proteins are annotated based on current literature as they relate to a biological pathway, process, function, or localization which is condensed within databases. Pathway analysis can identify whether these annotations are enriched within a set of genes or proteins, in this case the set of biomarkers reported within the eligible studies. Any biomarker reported to be significantly associated with or predictive of sPTB in at least one study were analyzed with gProfiler for enrichment using the Gene Ontology (GO) database for cellular component, and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database.

Results

Summary of search results

Search of the four databases retrieved 2002 non-duplicate records for screening. Screening of title and abstract identified 1695 records that were irrelevant to the research question and thus not further assessed for eligibility. A remaining 307 articles were assessed for eligibility using full-text records (Fig 1). The most common reason for record exclusion was study type, specifically review articles or conference abstracts (n = 91). Reference lists of review papers were reviewed, although no additional non-duplicate studies were identified. The remaining records were excluded by failing to meet the eligibility criteria for population (n = 23 studies measured biomarkers following the onset of labour symptoms, n = 33 did not meet criteria for healthy singleton pregnancies aged 18–35), exposure/marker (n = 1 amniotic fluid biomarker, n = 32 epigenetic/genetic biomarker), and outcome (n = 48 studies did not have sPTB as primary or secondary outcome). N = 66 and n = 76 studies identified by primary reviewer KKH and secondary reviewer EW, respectively, were cross-referenced by third reviewer DMS, leaving n = 79 studies included. The search was conducted again prior to submission to identify an additional n = 5 records, leaving a final total of n = 84 studies included for quality assessment.

thumbnail
Download:
Fig 1. PRISMA flow diagram.

Article identification, screening, and eligibility selection for systematic review of maternal blood markers associated with subsequent sPTB. Created with BioRender.com.

https://doi.org/10.1371/journal.pone.0265853.g001

Methodological quality assessment

Quality assessment scores for cohort studies ranged from 3 to 7 out of a total possible 7 points (S2 File). Areas with the lowest scores among cohort studies were assessment of outcome and adequacy of follow up. Scores for case-control studies ranged from 3 to 9 out of a total possible 9 points. Areas with the lowest scores among case-control studies were adequate case definitions, representativeness of cases, and comparability of cases and controls. Of the 84 studies identified as eligible, n = 7 studies did not reach 50% on the NOS score and were excluded from subsequent data extraction, leaving a remaining n = 77 studies included in the final review (Table 1).

thumbnail
Download:
Table 1. Summary and quality assessment of studies included in review.

https://doi.org/10.1371/journal.pone.0265853.t001

Study characteristics

Data extraction was performed on the eligible n = 77 papers. Studies included 48 case-control designs and 29 cohort studies. Study details including participant number, country of origin, methods and results are outlined in S3 File. A total of n = 278 unique biomarkers were identified as significantly associated with or predictive of sPTB in at least one study (S4 File).

Prediction models for preterm birth

Analysis methods from the 77 studies included, in order of complexity, univariate analysis comparing biomarker levels in sPTB pregnancies compared to term pregnancies (e.g., student’s t-test), binary classification (e.g., receiver operating characteristic curve analysis), multivariate models (e.g., linear or logistic regression), and machine learning models (e.g., random forest). Table 2 outlines the results from the 25 studies that conducted prediction of preterm birth using maternal blood biomarkers, as described by the sensitivity, specificity, and AUC of the classification. The AUC or area under the receiver operator curve is a performance metric for classification models, an AUC of 1.0 represents a perfect classifier, whereas an AUC of 0.5 represents random classification and thus is not a useful model. Top performing models included biomarkers identified through proteomic investigation, A2MG, HEMO, MBL2 [24], and ITIH4 [31], with an AUC of 0.89.

thumbnail
Download:
Table 2. Top prediction models for sPTB.

https://doi.org/10.1371/journal.pone.0265853.t002

Enrichment analysis of biomarkers

Based on gene ontology analysis for cellular component, 47 cellular component GO terms were enriched within the dataset. Nine of the top ten enriched terms are nested within the ‘extracellular region’ GO term, indicating there is significant enrichment of biomarkers localized to the extracellular space. Enrichment analysis using the HPA database found significant enrichment of biomarkers originating from placental syncytiotrophoblast cell bodies (p-adj = 0.006). A total 53 KEGG pathways were significantly enriched within the list of biomarkers. Of these, 32 pathways within the KEGG class ‘Human Diseases’ were excluded, as they were deemed not relevant to the physiology of labour and preterm labour. The remaining 21 enriched pathways (Table 3) were classed as ‘Immune’, ‘Signal Transduction’, ‘Signaling molecules and interaction’ ‘Development and regeneration’ and ‘Cell growth and death’.

thumbnail
Download:
Table 3. Enriched pathways within the set of biomarkers.

https://doi.org/10.1371/journal.pone.0265853.t003

Biomarkers from maternal serum screen tests

The most common biomarkers identified and analyzed were those measured during routine first and second trimester screening for aneuploidy, including pregnancy associated plasma protein A (PAPP-A), human chorionic gonadotropin (hCG), alpha fetoprotein (AFP), and estriol. Detailed odds ratios (OR) for maternal serum screen biomarkers summarized in S5 File.

PAPP-A.

Low levels of PAPP-A were associated with sPTB (OR range 1.7–5.4) [20, 36, 44, 45, 49–51, 64, 78, 89], though results were mixed [14, 42, 66, 69, 70].

hCG.

Evidence suggesting an association between maternal levels of hCG and sPTB is inconclusive [14, 20, 44, 50], though some studies suggest that low levels of hCG were associated with increased risk of sPTB (OR range 0.8–2.0) [43, 45, 49, 78] and that high levels of hCG independently decreased risk of sPTB [77].

AFP.

High serum AFP was significantly associated with sPTB (OR range 1.9–8.3) [14, 19, 35, 43–45, 58, 90] (OR range 1.9–8.3), though one study showed no significant difference [81].

Estriol.

Studies on serum estriol were split, while two studies showed a significant association between high estriol and sPTB [43, 61], others found no association [14, 44, 45].

Endocrine markers

CRH.

Elevated levels of corticotropin releasing hormone (CRH) were significantly associated with increased risk of sPTB [52, 71, 74]. Others reported elevated odds of sPTB with elevated levels of CRH, though the findings were not statistically significant [35, 58] (S5 File).

Cortisol.

Serum cortisol levels were not associated with gestational age at delivery [18], nor was elevated cortisol associated with increased odds of sPTB [35].

AMH.

Second trimester levels of anti-müllerian hormone (AMH) were not associated with sPTB: however, stable or rising levels of AMH in early pregnancy were associated with sPTB, but only in those with high levels of serum AFP [79].

Lipids and biomarkers associated with lipid pathways

Cholesterol.

Total cholesterol (TC) at the first trimester, as well as change in TC from the first to second trimesters, but not high density lipoproteins (HDL), low density lipoproteins (LDL) or triglyceride levels, improved prediction of sPTB compared to sPTB history alone, and performed similarly in those that did not have history of sPTB [14].

NEFAs.

High maternal blood levels of non-esterified or “free” fatty acids (NEFA) (>0.33mmol/L) was associated with 2.02 (confidence interval or CI 1.13–3.48) increased odds of subsequent sPTB compared to low NEFA levels (<0.19mmol/L) [25].

Eicosanoids and other lipid markers.

Biomarkers in the lipoxygenase, epoxygenase and cyclooxygenase pathways were investigated as potential biomarkers of sPTB. Lipid biomarkers alone performed similarly (AUC 0.79 CI 0.62–0.96) than a combined panel of inflammatory, oxidative stress and lipid biomarkers (AUC 0.79 CI 0.61–0.98) in distinguishing sPTB [16] from term births. Low levels of fatty acids eicosapentanaeoic acid (EPA) and docosahexaenoic acid (DHA) in the first and second trimesters was associated with 10-times increased risk of sPTB [60].

Heterogeneity of sPTB

Ethnicity-specific biomarkers.

The most common country of study origin was the United States of America, with majority non-Hispanic Caucasian or Black populations (S3 File). Corticotropin releasing hormone (CRH) was associated with sPTB in a majority ethnic Chinese population [52] and majority non-Hispanic Caucasian population [71], but not in a majority Black population [74]. Serum ferritin was associated with sPTB in Indian and Egyptian populations [12, 81], though results are mixed for Caucasian and Black populations in American and European countries [21, 31, 35, 63], which may suggest socio-demographic and ethnicity specific interactions within this biomarker. There is some suggestion that cortisol is associated with sPTB in Caucasian, but not Black populations [18, 35].

Parity.

The predictive value of alpha-fetoprotein (AFP) and multi-marker models was higher in parous populations as compared to nulliparous [14, 19, 57], indicating that these two populations may have distinct physiology of sPTB and may require distinct approaches to prediction. Other biomarkers from maternal serum screen tests, pregnancy associated plasma protein (PAPP-A) and human chorionic gonadotropin (B-hCG) were similarly more associated with sPTB in parous populations with previous sPTB as compared to nulliparous populations [20, 77].

BMI.

Inflammatory cytokines and CRP are more strongly associated with sPTB in populations with high BMI [23, 39], and that prediction using inflammatory biomarkers may be distinct in underweight and obese populations [28, 72].

Fetal sex-specific biomarkers.

There is some suggestion that there are no fetal sex-specific differences in biomarker associations with sPTB [15, 58, 79]. However, one 2010 study found that low pregnancy associated plasma protein (PAPP-A) levels were more strongly associated with sPTB in pregnancies carrying a female fetus as compared to male [49].

Gestational age considerations

Inflammatory biomarkers including cytokines and serum ferritin, biomarkers from maternal serum screen tests and CRH are more strongly associated with very early sPTB (<32 weeks gestation) as compared to moderate-late sPTB [34, 35, 45, 46, 48, 52, 76]. On the other hand, gene expression markers, and protein markers pro-MBP and SP1 may be more associated with overall sPTB (<37 weeks gestation) as compared to early [55, 66, 67], while estriol is more strongly associated with moderate sPTB (32–34 weeks) [61]. A 2015 study of Tanzanian women found distinct angiogenic biomarkers for each subtype of sPTB prematurity, indicating distinct physiologies across subtypes [91].

Of the 77 studies included in this review, only 10 reported biomarker measurements obtained at more than one time point. Three studies on transcript markers found that change in transcript levels was more predictive than absolute values [38, 80, 87]. On the other hand, Parry et al. [62], showed that change in proteomic markers was not highly associated with sPTB. Five additional studies did not investigate change in biomarker levels, but performed statistical analysis at each time point separately [15, 26, 33, 34, 37], while Esplin et al., [31] measured markers in separate cohorts for each time point. This limits our understanding of how biomarker levels change throughout gestation, which may be an important indicator for risk of sPTB. Inflammatory cytokines showed no difference in association when measured at different timepoints in gestation, except for IL10, which was most strongly associated with sPTB when measured after 22 weeks gestation [33, 34]. A 2010 study found that thrombin-antithrombin complexes (TAT) were more strongly associated with sPTB when measured later in gestation (28 weeks) as compared to earlier (24 weeks) [37]. On the other hand, a 2011 study found the free B-hCG was more strongly associated with sPTB with earlier sampling, but the same difference was not true of PAPP-A [50]. One study compared the predictive value at multiple time points and found that among a range of time points between 17–28 weeks, samples collected at 19–21 weeks were ideal for biomarker discovery [72].

Discussion

Main findings

Analysis of the seventy-seven papers identified via this systematic review suggest that there is no clear single biomarker or set of biomarkers in the current existing literature for the prediction of sPTB. Low levels of PAPP-A and elevated levels of AFP are associated with increased risk of sPTB; however, these results may be biased due to secondary use of data and incomplete datasets. Consistent study design, which would facilitate systematic meta-analyses of these studies, would be necessary to validate these results before confirming any clinical utility of these markers. Further, although inflammation has long been associated with labour and sPTB, studies investigating the association between inflammatory biomarkers and sPTB are inconclusive. G-CSF was found to have the strongest and most consistent association with sPTB, but there is insufficient evidence to support an association with other biomarkers associated with systemic inflammation such as IL-1B, TNFa, CRP, IL-6, IL-2, IFNy, IL-10 and serum ferritin. However, the emergence of omics technologies has identified biomarkers and pathways of interest that may identify novel avenues for prediction. It is likely that no single biomarker will be predictive of sPTB but high-throughput technologies for biomarker discovery, improved feature selection, and integration with other known risk factors such as cervical length and history of sPTB may provide a set of biomarkers with clinical utility for the prediction of sPTB.

The preterm birth phenotype

This systematic review was the first, to our knowledge, to systematically review the existing literature on maternal blood markers, collected before any signs of labour, that are predictive of a spontaneous preterm birth. From our initial search, forty-eight primary studies were excluded for not defining preterm birth following spontaneous labour or rupture of membranes as their primary or secondary outcome(s). Outcomes were typically preterm delivery (<37 weeks), with no reference to primary records that would differentiate the spontaneous PTBs from those that follow physician initiation, otherwise referred to as medically indicated PTBs. These medically indicated births and those arising spontaneously are not likely to have a shared etiology, and thus identifying common biomarkers for all PTB subtypes is unlikely; this was the basis of our justification to exclude these papers from review.

A limitation of the studies reviewed is the potential for misclassification bias due to poor outcome definitions. Most studies did not provide reference to primary records that define the sPTB outcome. With respect to gestational age, misclassification bias is most likely for those preterm cases that occur near term (~36 weeks), depending on the method by which gestational age was determined (last menstrual period, ultrasound etc.). With respect to other obstetric outcomes, most studies failed to provide repeatable protocols on how sPTBs were differentiated from those that were medically indicated, and the extent of missing data may have affected the risk of bias within the reported results.

While preterm birth is often regarded as a single outcome in clinical practice, the preterm birth phenotype has multiple and complex etiologies. There is no recognized system for grouping preterm birth phenotypes, and the known etiologies include pre-eclampsia, multiple births, infection, fetal growth restriction, fetal distress, decidual hemorrhage and placental dysfunction [92]. However, 30% of preterm births, a greater proportion than any other etiology, are not associated with known maternal, fetal, or placental conditions, but exhibit spontaneous contractions or rupture of membranes [93]. Increased fetal mortality and morbidity is not only associated with lower gestational age at delivery, but also with the different patterns of PTB etiology [94]. For example, it is likely that extreme PTBs (<28 weeks gestation) have a different etiology than those occurring near term. Larger studies that allow for a high degree of stratification, such as by gestational age (extreme PTB <28 weeks, very PTB 28–32 weeks, or moderate-late PTB 32–37 weeks), and by potential etiology may identify novel biomarkers unique to each phenotype and may mediate the effects of the heterogeneity. Biomarker discovery for early detection of highly heterogenous outcomes requires at least 2-fold larger sample sizes and different statistical considerations than if the outcome were homogenous, which must be put into consideration when performing power calculations during study design [95].

Considerations for sample collection

Our results suggest that change in biomarkers over multiple measurements throughout gestation is more strongly associated with sPTB than single timepoint measurements, though this may be dependent on the type of biomarker measured. Not only is there evidence to suggest that multiple measurements are beneficial for biomarker discovery and predicting sPTB, understanding the dynamics of molecular changes throughout gestation would also provide greater insight to the mechanisms of sPTB. Sample collection in large cohorts is time consuming and costly, not excluding the time for additional analyses, which is further exacerbated by multiple measurements. While there is evidence to suggest that two measurements are beneficial for predicting sPTB, there is insufficient evidence to suggest that more than two would provide any added benefit. Further, our results demonstrate that biomarkers measured in the second and third trimesters are more predictive of sPTB than those measured earlier, though this likely depends on the biomarker. Results of a chi-squared analysis of markers measured in both serum and plasma found no statistical difference in the likelihood of reporting an association with sPTB (p>0.05). There is insufficient evidence to suggest an advantage to collecting plasmas over serums, and as serums are often collected as part of routine antenatal screening, this may be a more convenient biological tissue to use for biomarker discovery. However, a direct comparison of biomarkers measured in both serums and plasmas would be necessary to investigate the use of either biofluid.

The ‘Big Data’ era

High throughput technologies such as microarrays, next generation sequencing and mass spectrometry now allow for the generation of large datasets including proteomic, metabolomic, genomic and transcriptomic information. These technologies, along with subsequent bioinformatic analyses, allows for an unprecedented opportunity to identify novel biomarkers [96]. Investigations into maternal biomarkers for sPTB are often limited by the currently limited understanding of the molecular mechanisms of labour. For example, though there is an ongoing hypothesis that labour is an inflammatory process [97], we show here that biomarkers of systemic inflammation such as IL-1B and CRP are not associated with sPTB. The process of labour may certainly involve inflammatory mediators but is likely much more complex than a simple switching on or off, of inflammation. These large omics-datasets, while identifying novel markers, also allow for investigation into global molecular changes, providing additional insight into novel pathways and molecules that may be involved in the mechanism of preterm birth. In particular, the field of metabolomics is largely under studied with respect to prediction of sPTB, only one study met the inclusion criteria for this review [26]. The field of parturition requires further investigation using these high-throughput technologies to facilitate identification of better biomarkers with clinical utility for prediction of preterm birth.

With the advent of such high-throughput technologies and increasingly high-dimensional datasets comes the need for more robust forms of data analysis and mining. Machine learning and other complex data analysis methods are particularly well suited for high-dimensional and complex data as they do not generally require the data to adhere to any a priori assumptions about linearity of distribution [98]. These techniques can be used to identify subtle, complex and interactive patterns within datasets which can subsequently be leveraged for prediction and discovery of phenotypes such as sPTB [96]. A major limitation to machine learning is the risk of overfitting the data, which is especially problematic when the number of features far exceeds the number of observations, as is common in -omics datasets in health sciences. This drives the impetus for dimensionality reduction and/or feature selection, which can reduce the risk of overfitting, reduce computational resources required for analysis and, in the case of supervised feature selection, can identify the most important features for prediction. A recent study identified intra- and extra-uterine factors most informative of sPTB to estimate risk of sPTB using a random forest classifier with high predictive performance (AUC 0.81) [99]. The etiology of sPTB is multi-factorial, and is likely driven by more complex, subtle interactions which machine learning approaches, along with feature selection for identifying informative features, is well suited to detect.

Limitations

This review was not preregistered. Pre-registering systematic reviews can reduce potential for bias, increase transparency and avoid unintended duplication of reviews [100]. Though there have been recently published systematic reviews on the topic of biomarkers of PTB in 2017 and 2018 [101, 102], there has not, to our knowledge, been a systematic review of biomarkers specific to the maternal blood compartment prior to the onset of labour. Further, studies published within the last 5 years since their publication, particularly those using high-throughput proteomics and transcriptomic techniques have identified novel biomarkers of sPTB, highlighting the value of revisiting the literature. A 2002 study of 47 systematic reviews found that 91.5% of Cochrane reviews contained major changes to their methods and selected outcomes as compared to their pre-registered protocols [103], suggesting a potential risk of bias in the reported outcomes.

Another limitation of this study is that we did not collect grey literature (e.g. unpublished work). Non reporting biases contribute to missing results in this systematic review, likely skewing associations with sPTB more positively. Negative results, in other words, biomarkers that have been found to not be predictive of sPTB, are less likely to be published due to high p value, the magnitude or direction of the results. Biomarkers that did not meet thresholds for statistical significance in preliminary/bivariate or univariate analysis were often excluded from downstream analyses, or the data was not shown, suggesting that there was further under-reporting of results due to insufficient p-values. Further, support for why the biomarkers of interest were selected in each study was often missing, not thorough, or not compelling. Studies that prospectively register their protocols are more likely to report negative results [104]; however, observational studies, which make up the majority of studies included in this review, are not typically prospectively registered, limiting options to reduce bias due to missing results. Lastly, we did not perform a meta-analysis, primarily due to inconsistent study designs which preclude options for meta-analyses but also that a meta-analysis of biomarkers assessed in the literature would not be particularly useful as it would be so heavily biased by the authors’ choice of biomarker—apart from, perhaps, -omics studies. Meta-analysis of the -omics studies was limited by highly inconsistent study designs and thus was not conducted.

Conclusions

Currently, there is no known clear single biomarker, or set of biomarkers, for the prediction of spontaneous preterm birth. This review highlights that current biomarker discovery techniques are largely limited by the heterogenous nature of preterm birth and an incomplete understanding of the mechanisms that drive this process. Omics-style studies with more robust feature selection and analytical approaches provide a promising avenue for the identification of novel biomarkers. Larger studies with adequate power and more consistent study design, namely, clearly defined outcomes that consider the heterogeneity and subtypes of PTB, are needed to identify a set of biomarkers predictive of sPTB.

Acknowledgments

The authors would like to thank Helen Lee Robertson and the University of Calgary Health Science Library for assistance with the database searches.

References

  1. 1. Gyamfi-Bannerman C, Ananth CV. Trends in Spontaneous and Indicated Preterm Delivery Among Singleton Gestations in the United States, 2005–2012. OBSTET GYNECOL. 2014;124(6):1069–74. pmid:25415156
  2. 2. Berghella V, Saccone G. Fetal fibronectin testing for reducing the risk of preterm birth. Cochrane Database of Systematic Reviews. 2019;7:CD006843. pmid:31356681
  3. 3. Dos Santos F, Daru J, Rogozińska E, Cooper NAM. Accuracy of fetal fibronectin for assessing preterm birth risk in asymptomatic pregnant women: a systematic review and meta-analysis. ACTA OBSTET GYN SCAN. 2018;97(6):657–67. pmid:29355887
  4. 4. Phillips C, Velji Z, Hanly C, Metcalfe A. Risk of recurrent spontaneous preterm birth: a systematic review and meta-analysis. BMJ OPEN. 2017;7(6):e015402–e. pmid:28679674
  5. 5. Yuan M, Jordan F, McInnes IB, Harnett MM, Norman JE. Leukocytes are primed in peripheral blood for activation during term and preterm labour. MOL HUM REPROD. 2009;15(11):713–24. pmid:19628509
  6. 6. Wei SQ, Fraser W, Luo ZC. Inflammatory cytokines and spontaneous preterm birth in asymptomatic women: A systematic review. Obstetrics and Gynecology. 2010;116(2 PART 1):393–401. pmid:20664401
  7. 7. Morris RK, Meller CH, Tamblyn J, Malin GM, Riley RD, Kilby MD, et al. Association and prediction of amniotic fluid measurements for adverse pregnancy outcome: systematic review and meta-analysis. BJOG-INT J OBSTET GY. 2014;121(6):686–99. pmid:24738894
  8. 8. Lee SY, Park KH, Jeong EH, Oh KJ, Ryu A, Kim A. Intra-amniotic infection/inflammation as a risk factor for subsequent ruptured membranes after clinically indicated amniocentesis in preterm labor. J KOREAN MED SCI. 2013;28(8):1226–32. pmid:23960452
  9. 9. Cao L. Data Science: A Comprehensive Overview. ACM Comput Surv. 2017;50(3):Article 43.
  10. 10. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med. 2009;151(4):264–9, w64. pmid:19622511
  11. 11. García-Campos MA, Espinal-Enríquez J, Hernández-Lemus E. Pathway Analysis: State of the Art. Frontiers in Physiology. 2015;6. pmid:26733877
  12. 12. Abdel-Malek K, El-Halwagi MA, Hammad BE, Azmy O, Helal O, Eid M, et al. Role of maternal serum ferritin in prediction of preterm labour. Journal of Obstetrics & Gynaecology. 2018;38(2):222–5. pmid:28903602
  13. 13. Akoto C, Chan CYS, Tshivuila-Matala COO, Ravi K, Zhang W, Vatish M, et al. Innate lymphoid cells are reduced in pregnant HIV positive women and are associated with preterm birth. SCI REP-UK. 2020;10(1):13265-. pmid:32764636
  14. 14. Alleman BW, Smith AR, Byers HM, Bedell B, Ryckman KK, Murray JC, et al. A proposed method to predict preterm birth using clinical data, standard maternal serum screening, and cholesterol. American Journal of Obstetrics and Gynecology. 2013;208(6):472.e1–.e11.
  15. 15. Ashrap P, Watkins DJ, Mukherjee B, Boss J, Richards MJ, Rosario Z, et al. Maternal blood metal and metalloid concentrations in association with birth outcomes in Northern Puerto Rico. Environment International. 2020;138. pmid:32179314
  16. 16. Aung MT, Yu Y, Ferguson KK, Cantonwine DE, Zeng L, McElrath TF, et al. Prediction and associations of preterm birth and its subtypes with eicosanoid enzymatic pathways and inflammatory markers. Scientific Reports. 2019;9(1):17049. pmid:31745121
  17. 17. Bakalis SP, Poon LCY, Vayna AM, Pafilis I, Nicolaides KH. C-reactive protein at 11–13 weeks’ gestation in spontaneous early preterm delivery. Journal of Maternal-Fetal and Neonatal Medicine. 2012;25(12):2475–8. pmid:22900797
  18. 18. Bandoli G, Jelliffe-Pawlowski LL, Feuer SK, Liang L, Oltman SP, Paynter R, et al. Second trimester serum cortisol and preterm birth: an analysis by timing and subtype. Journal of Perinatology. 2018;38(8):973–81. pmid:29795321
  19. 19. Beta J, Bredaki FE, Rodriguez Calvo J, Akolekar R, Nicolaides KH. Maternal serum α-fetoprotein at 11–13 weeks’ gestation in spontaneous early preterm delivery. Fetal Diagnosis and Therapy. 2011;30(2):88–93. pmid:21411977
  20. 20. Beta J, Akolekar R, Ventura W, Syngelaki A, Nicolaides KH. Prediction of spontaneous preterm delivery from maternal factors, obstetric history and placental perfusion and function at 11–13 weeks. Prenatal Diagnosis. 2011;31(1):75–83. pmid:21210482
  21. 21. Beta J, Poon LCY, Bakalis S, Mosimann B, Nicolaides KH. Maternal serum ferritin at 11- to 13-week gestation in spontaneous early preterm delivery. Journal of Maternal-Fetal and Neonatal Medicine. 2012;25(10):1852–5. pmid:22468901
  22. 22. Bradford C, Severinsen R, Pugmire T, Rasmussen M, Stoddard K, Uemura Y, et al. Analytical validation of protein biomarkers for risk of spontaneous preterm birth. Clinical Mass Spectrometry. 2017;3:25–38.
  23. 23. Bullen BL, Jones NM, Holzman CB, Tian Y, Senagore PK, Thorsen P, et al. C-reactive protein and preterm delivery: Clues from placental findings and maternal weight. Reproductive Sciences. 2013;20(6):715–22. pmid:23221172
  24. 24. Cantonwine DE, Zhang Z, Rosenblatt K, Goudy KS, Doss RC, Ezrin AM, et al. Evaluation of proteomic biomarkers associated with circulating microparticles as an effective means to stratify the risk of spontaneous preterm birth. American Journal of Obstetrics and Gynecology. 2016;214(5):631.e1–.e11. pmid:26874302
  25. 25. Catov JM, Bertolet M, Chen YF, Evans RW, Hubel CA. Nonesterified fatty acids and spontaneous preterm birth: A factor analysis for identification of risk patterns. American Journal of Epidemiology. 2014;179(10):1208–15. pmid:24714724
  26. 26. Considine EC, Khashan AS, Kenny LC. Screening for preterm birth: Potential for a metabolomics biomarker panel. Metabolites. 2019;9(5). pmid:31067710
  27. 27. Curry AE, Vogel I, Drews C, Schendel D, Skogstrand K, Flanders WD, et al. Mid-pregnancy maternal plasma levels of interleukin 2, 6, and 12, tumor necrosis factor-alpha, interferon-gamma, and granulocyte-macrophage colony-stimulating factor and spontaneous preterm delivery. Acta Obstetricia et Gynecologica Scandinavica. 2007;86(9):1103–10. pmid:17712652
  28. 28. Curry AE, Thorsen P, Drews C, Schendel D, Skogstrand K, Flanders WD, et al. First-trimester maternal plasma cytokine levels, pre-pregnancy body mass index, and spontaneous preterm delivery. Acta Obstetricia et Gynecologica Scandinavica. 2009;88(3):332–42. pmid:19241227
  29. 29. Dhaifalah I, Andrys C, Drahosova M, Musilova I, Adamik Z, Kacerovsky M. Azurocidin levels in maternal serum in the first trimester can predict preterm prelabor rupture of membranes. J Matern Fetal Neonatal Med. 2014;27(5):511–5. pmid:23808364
  30. 30. El-Achi V, de Vries B, O’Brien C, Park F, Tooher J, Hyett J. First-Trimester Prediction of Preterm Prelabour Rupture of Membranes. FETAL DIAGN THER. 2020;47(8):624–9. pmid:32268324
  31. 31. Esplin MS, Merrell K, Goldenberg R, Lai Y, Iams JD, Mercer B, et al. Proteomic identification of serum peptides predicting subsequent spontaneous preterm birth. American Journal of Obstetrics and Gynecology. 2011;204(5):391.e1–.e8. pmid:21074133
  32. 32. Ezrin AM, Brohman B, Willmot J, Baxter S, Moore K, Luther M, et al. Circulating serum-derived microparticles provide novel proteomic biomarkers of spontaneous preterm birth. Am J Perinatol. 2015;32(6):605–14. pmid:25825961
  33. 33. Ferguson KK, McElrath TF, Chen YH, Mukherjee B, Meeker JD. Longitudinal profiling of inflammatory cytokines and C-reactive protein during uncomplicated and preterm pregnancy. American Journal of Reproductive Immunology. 2014;72(3):326–36. pmid:24807462
  34. 34. Goldenberg RL, Andrews WW, Mercer BM, Moawad AH, Meis PJ, Iams JD, et al. The preterm prediction study: granulocyte colony-stimulating factor and spontaneous preterm birth. National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. American Journal of Obstetrics & Gynecology. 2000;182(3):625–30. pmid:10739519
  35. 35. Goldenberg RL, Iams JD, Mercer BM, Meis PJ, Moawad A, Das A, et al. The Preterm Prediction Study: Toward a multiple-marker test for spontaneous preterm birth. American Journal of Obstetrics and Gynecology. 2001;185(3):643–51. pmid:11568793
  36. 36. Gupta S, Goyal M, Verma D, Sharma A, Bharadwaj N, Kabra M, et al. Adverse pregnancy outcome in patients with low pregnancy-associated plasma protein-A: The Indian Experience. Journal of Obstetrics & Gynaecology Research. 2015;41(7):1003–8. pmid:25773764
  37. 37. Hackney DN, Catov JM, Simhan HN. Low concentrations of thrombin-inhibitor complexes and the risk of preterm delivery. American Journal of Obstetrics & Gynecology. 2010;203(2):184.e1–6. pmid:20510913
  38. 38. Heng YJ, Pennell CE, McDonald SW, Vinturache AE, Xu J, Lee MWF, et al. Maternal whole blood gene expression at 18 and 28 weeks of gestation associated with spontaneous preterm birth in asymptomatic women. PLoS ONE. 2016;11(6). pmid:27333071
  39. 39. Huang L, Hou Q, Huang Y, Ye J, Huang S, Tian J, et al. Serum multiple cytokines for the prediction of spontaneous preterm birth in asymptomatic women: A nested case-control study. Cytokine. 2019;117:91–7. pmid:30831445
  40. 40. Huang S, Tian J, Liu C, Long Y, Cao D, Wei L, et al. Elevated C-reactive protein and complement C3 levels are associated with preterm birth: A nested case-control study in Chinese women. BMC Pregnancy and Childbirth. 2020;20(1). pmid:32106828
  41. 41. Hvilsom GB, Thorsen P, Jeune B, Bakketeig LS. C-reactive protein: a serological marker for preterm delivery? Acta Obstetricia et Gynecologica Scandinavica. 2002;81(5):424–9. pmid:12027816
  42. 42. Inan C, Varol FG, Erzincan SG, Uzun I, Sutcu H, Sayin NC. Use of prokineticin-1 (PROK1), pregnancy-associated plasma protein A (PAPP-A) and PROK1/PAPP-A ratio to predict adverse pregnancy outcomes in the first trimester: a prospective study. Journal of Maternal-Fetal and Neonatal Medicine. 2018;31(20):2685–92. pmid:28675948
  43. 43. Jelliffe-Pawlowski LL, Baer RJ, Currier RJ. Second trimester serum predictors of preterm birth in a population-based sample of low-risk pregnancies. Prenatal Diagnosis. 2010;30(8):727–33. pmid:20661885
  44. 44. Jelliffe-Pawlowski LL, Shaw GM, Currier RJ, Stevenson DK, Baer RJ, O’Brodovich HM, et al. Association of early-preterm birth with abnormal levels of routinely collected first- and second-trimester biomarkers. American Journal of Obstetrics and Gynecology. 2013;208(6):492.e1–.e11. pmid:23395922
  45. 45. Jelliffe-Pawlowski LL, Baer RJ, Blumenfeld YJ, Ryckman KK, O’Brodovich HM, Gould JB, et al. Maternal characteristics and mid-pregnancy serum biomarkers as risk factors for subtypes of preterm birth. BJOG: An International Journal of Obstetrics and Gynaecology. 2015;122(11):1484–93. pmid:26111589
  46. 46. Jelliffe-Pawlowski LL, Rand L, Bedell B, Baer RJ, Oltman SP, Norton ME, et al. Prediction of preterm birth with and without preeclampsia using mid-pregnancy immune and growth-related molecular factors and maternal characteristics. Journal of Perinatology. 2018;38(8):963–72. pmid:29795450
  47. 47. Kansu-Celik H, Tasci Y, Karakaya BK, Cinar M, Candar T, Caglar GS. Maternal serum advanced glycation end products level as an early marker for predicting preterm labor/PPROM: a prospective preliminary study. Journal of Maternal-Fetal and Neonatal Medicine. 2019;32(16):2758–62. pmid:29506420
  48. 48. Khambalia AZ, Collins CE, Roberts CL, Morris JM, Powell KL, Tasevski V, et al. High maternal serum ferritin in early pregnancy and risk of spontaneous preterm birth. British Journal of Nutrition. 2015;114(3):455–61. pmid:26146276
  49. 49. Kirkegaard I, Uldbjerg N, Petersen OB, Torring N, Henriksen TB. PAPP-A, free beta-hCG, and early fetal growth identify two pathways leading to preterm delivery. Prenatal Diagnosis. 2010;30(10):956–63. pmid:20721873
  50. 50. Kirkegaard I, Henriksen TB, Torring N, Uldbjerg N. PAPP-A and free beta-hCG measured prior to 10 weeks is associated with preterm delivery and small-for-gestational-age infants. Prenatal Diagnosis. 2011;31(2):171–5. pmid:21268036
  51. 51. Kwik M, Morris J. Association between firsttrimester maternal serum pregnancy associated plasma protein-A and adverse pregnancy outcome. Australian & New Zealand Journal of Obstetrics & Gynaecology. 2003;43(6):438–42.
  52. 52. Leung TN, Chung TK, Madsen G, McLean M, Chang AM, Smith R. Elevated mid-trimester maternal corticotrophin-releasing hormone levels in pregnancies that delivered before 34 weeks. British Journal of Obstetrics & Gynaecology. 1999;106(10):1041–6. pmid:10519429
  53. 53. Lynch AM, Wagner BD, Deterding RR, Giclas PC, Gibbs RS, Janoff EN, et al. The relationship of circulating proteins in early pregnancy with preterm birth. American Journal of Obstetrics and Gynecology. 2016;214(4):517.e1–.e8. pmid:26576488
  54. 54. Ma M, Zhu M, Zhuo B, Li L, Chen H, Xu L, et al. Use of complete blood count for predicting preterm birth in asymptomatic pregnant women: A propensity score-matched analysis. J CLIN LAB ANAL. 2020;34(8):e23313–n/a. pmid:32222021
  55. 55. Manuck TA, Eaves LA, Rager JE, Fry RC. Mid-pregnancy maternal blood nitric oxide-related gene and miRNA expression are associated with preterm birth. Epigenomics. 2021;13(9):667–82. pmid:33890487
  56. 56. McDonald CR, Darling AM, Conroy AL, Tran V, Cabrera A, Liles WC, et al. Inflammatory and angiogenic factors at mid-pregnancy are associated with spontaneous preterm birth in a cohort of Tanzanian women. PLoS ONE. 2015;10(8).
  57. 57. McElrath TF, Cantonwine DE, Jeyabalan A, Doss RC, Page G, Roberts JM, et al. Circulating microparticle proteins obtained in the late first trimester predict spontaneous preterm birth at less than 35 weeks’ gestation: a panel validation with specific characterization by parity. AM J OBSTET GYNECOL. 2019;220(5):488.e1–.e11. pmid:30690014
  58. 58. McLean M, Bisits A, Davies J, Walters W, Hackshaw A, De Voss K, et al. Predicting risk of preterm delivery by second-trimester measurement of maternal plasma corticotropin-releasing hormone and alpha-fetoprotein concentrations. American Journal of Obstetrics & Gynecology. 1999;181(1):207–15. pmid:10411821
  59. 59. Ngo TTM, Moufarrej MN, Rasmussen MLH, Camunas-Soler J, Pan W, Okamoto J, et al. Noninvasive blood tests for fetal development predict gestational age and preterm delivery. Science. 2018;360(6393):1133–6. pmid:29880692
  60. 60. Olsen SF, Halldorsson TI, Thorne-Lyman AL, Strom M, Gortz S, Granstrom C, et al. Plasma Concentrations of Long Chain N-3 Fatty Acids in Early and Mid-Pregnancy and Risk of Early Preterm Birth. EBioMedicine. 2018;35:325–33. pmid:30082226
  61. 61. Olsen RN, Dunsmoor-Su R, Capurro D, McMahon K, Gravett MG. Correlation between spontaneous preterm birth and mid-trimester maternal serum estriol. Journal of Maternal-Fetal and Neonatal Medicine. 2014;27(4):376–80. pmid:23777262
  62. 62. Parry S, Zhang H, Biggio J, Bukowski R, Varner M, Xu Y, et al. Maternal serum serpin B7 is associated with early spontaneous preterm birth. American Journal of Obstetrics and Gynecology. 2014;211(6):678.e1–.e12. pmid:24954659
  63. 63. Paternoster DM, Stella A, Gerace P, Manganelli F, Plebani M, Snijders D, et al. Biochemical markers for the prediction of spontaneous pre-term birth. International Journal of Gynaecology & Obstetrics. 2002;79(2):123–9. pmid:12427396
  64. 64. Patil M, Panchanadikar TM, Wagh G. Variation of papp-a level in the first trimester of pregnancy and its clinical outcome. Journal of Obstetrics & Gynaecology of India. 2014;64(2):116–9. pmid:24757339
  65. 65. Petersen LK, Skajaa K, Uldbjerg N. Serum relaxin as a potential marker for preterm labour. British Journal of Obstetrics & Gynaecology. 1992;99(4):292–5. pmid:1581273
  66. 66. Pihl K, Larsen T, Laursen I, Krebs L, Christiansen M. First trimester maternal serum pregnancy-specific beta-1-glycoprotein (SP1) as a marker of adverse pregnancy outcome. Prenatal Diagnosis. 2009;29(13):1256–61. pmid:19911417
  67. 67. Pihl K, Larsen T, Rasmussen S, Krebs L, Christiansen M. The proform of eosinophil major basic protein: a new maternal serum marker for adverse pregnancy outcome. Prenatal Diagnosis. 2009;29(11):1013–9. pmid:19626619
  68. 68. Pitiphat W, Gillman MW, Joshipura KJ, Williams PL, Douglass CW, Rich-Edwards JW. Plasma C-reactive protein in early pregnancy and preterm delivery. American Journal of Epidemiology. 2005;162(11):1108–13. pmid:16236995
  69. 69. Poon LCY, Nekrasova E, Anastassopoulos P, Livanos P, Nicolaides KH. First-trimester maternal serum matrix metalloproteinase-9 (MMP-9) and adverse pregnancy outcome. Prenatal Diagnosis. 2009;29(6):553–9. pmid:19242924
  70. 70. Poon LCY, Musci T, Song K, Syngelaki A, Nicolaides KH. Maternal plasma cell-free fetal and maternal DNA at 11–13 weeks’ gestation: Relation to fetal and maternal characteristics and pregnancy outcomes. Fetal Diagnosis and Therapy. 2013;33(4):215–23. pmid:23466432
  71. 71. Ruiz RJ, Fullerton J, Brown CE, Dudley DJ. Predicting risk of preterm birth: the roles of stress, clinical risk factors, and corticotropin-releasing hormone. Biological Research for Nursing. 2002;4(1):54–64. pmid:12363283
  72. 72. Saade GR, Boggess KA, Sullivan SA, Markenson GR, Iams JD, Coonrod DV, et al. Development and validation of a spontaneous preterm delivery predictor in asymptomatic women. American Journal of Obstetrics and Gynecology. 2016;214(5):633e1–e24. pmid:26874297
  73. 73. Shin JE, Shin JC, Kim SJ, Lee Y, Park IY, Lee S. Early midtrimester serum insulin-like growth factors and cervical length to predict preterm delivery. Taiwanese Journal of Obstetrics and Gynecology. 2016;55(1):45–9. pmid:26927247
  74. 74. Sibai B, Meis PJ, Klebanoff M, Dombrowski MP, Weiner SJ, Moawad AH, et al. Plasma CRH measurement at 16 to 20 weeks’ gestation does not predict preterm delivery in women at high-risk for preterm delivery. American Journal of Obstetrics and Gynecology. 2005;193(3 SUPPL.):1181–6.
  75. 75. Smith GCS, Crossley JA, Aitken DA, Jenkins N, Lyall F, Cameron AD, et al. Circulating angiogenic factors in early pregnancy and the risk of preeclampsia, intrauterine growth restriction, spontaneous preterm birth, and stillbirth. Obstetrics and Gynecology. 2007;109(6):1316–24. pmid:17540803
  76. 76. Smith GCS, Shah I, Crossley JA, Aitken DA, Pell JP, Nelson SM, et al. Pregnancy-associated plasma protein A and alpha-fetoprotein and prediction of adverse perinatal outcome. Obstetrics and Gynecology. 2006;107(1):161–6. pmid:16394054
  77. 77. Soni S, Krantz DA, Blitz MJ, Vohra N, Rochelson B. Elevated maternal serum-free beta-human chorionic gonadotropin (beta-hCG) and reduced risk of spontaneous preterm delivery. Journal of Maternal-Fetal & Neonatal Medicine. 2018:1–6.
  78. 78. Spencer K, Cowans NJ, Molina F, Kagan KO, Nicolaides KH. First-trimester ultrasound and biochemical markers of aneuploidy and the prediction of preterm or early preterm delivery. Ultrasound in Obstetrics and Gynecology. 2008;31(2):147–52. pmid:17992705
  79. 79. Stegmann BJ, Santillan M, Leader B, Smith E, Santillan D. Changes in antimullerian hormone levels in early pregnancy are associated with preterm birth. Fertility and Sterility. 2015;104(2):347–55.e3. pmid:26074093
  80. 80. Tarca AL, Pataki BÁ, Romero R, Sirota M, Guan Y, Kutum R, et al. Crowdsourcing assessment of maternal blood multi-omics for predicting gestational age and preterm birth. Cell Reports Medicine. 2021;2(6):100323-. pmid:34195686
  81. 81. Tripathi R, Tyagi S, Singh N, Mala YM, Singh C, Bhalla P, et al. Can preterm labour be predicted in low risk pregnancies? Role of clinical, sonographic, and biochemical markers. Journal of Pregnancy. 2014;2014:623269. pmid:25405034
  82. 82. Vogel I, Goepfert AR, Moller HJ, Cliver S, Thorsen P, Andrews WW. Early mid-trimester serum relaxin, soluble CD163, and cervical length in women at high risk for preterm delivery. American Journal of Obstetrics and Gynecology. 2006;195(1):208–14. pmid:16600167
  83. 83. Vogel I, Goepfert AR, Thorsen P, Skogstrand K, Hougaard DM, Curry AH, et al. Early second-trimester inflammatory markers and short cervical length and the risk of recurrent preterm birth. Journal of Reproductive Immunology. 2007;75(2):133–40. pmid:17442403
  84. 84. Whitcomb BW, Schisterman EF, Luo X, Chegini N. Maternal serum granulocyte colony-stimulating factor levels and spontaneous preterm birth. Journal of Women’s Health. 2009;18(1):73–8. pmid:19105692
  85. 85. Winger EE, Reed JL, Ji X, Gomez-Lopez N, Pacora P, Romero R. MicroRNAs isolated from peripheral blood in the first trimester predict spontaneous preterm birth. PLoS ONE. 2020;15(8 August). pmid:32790689
  86. 86. Wommack JC, Trzeciakowski JP, Miranda RC, Stowe RP, Jeanne Ruiz R. Micro RNA clusters in maternal plasma are associated with preterm birth and infant outcomes. PLoS ONE. 2018;13(6). pmid:29949620
  87. 87. Zhou G, Holzman C, Chen B, Wang P, Heng YJ, Kibschull M, et al. EBF1-Correlated Long Non-coding RNA Transcript Levels in 3rd Trimester Maternal Blood and Risk of Spontaneous Preterm Birth. Reproductive Sciences. 2020. pmid:32959224
  88. 88. Zhu H, Yang MJ. Maternal plasma concentrations of macrophage migration inhibitory factor at first trimester as a predictive biomarker of preterm delivery in Chinese women. Clinica Chimica Acta. 2018;483:286–90. pmid:29684382
  89. 89. Pummara P, Tongsong T, Wanapirak C, Sirichotiyakul S, Luewan S. Association of first-trimester pregnancy-associated plasma protein A levels and idiopathic preterm delivery: A population-based screening study. Taiwanese Journal of Obstetrics & Gynecology. 2016;55(1):72–5. pmid:26927253
  90. 90. Smith GCS, Shah I, White IR, Pell JP, Crossley JA, Dobbie R. Maternal and biochemical predictors of spontaneous preterm birth among nulliparous women: A systematic analysis in relation to the degree of prematurity. International Journal of Epidemiology. 2006;35(5):1169–77. pmid:16882673
  91. 91. McDonald CE, Darling AM, Conroy AL, Hayford K, Rajwans N, Aboud S, et al. Imbalance of inflammatory and angiogenic factors in early pregnancy are associated with preterm birth in a prospective cohort of malariaexposed tanzanian women. American Journal of Tropical Medicine and Hygiene. 2013;89(5):456.
  92. 92. Frey HA, Klebanoff MA. The epidemiology, etiology, and costs of preterm birth. SEMIN FETAL NEONAT M. 2015;21(2):68–73.
  93. 93. Barros FC, Papageorghiou AT, Victora CG, Noble JA, Pang R, Iams J, et al. The Distribution of Clinical Phenotypes of Preterm Birth Syndrome: Implications for Prevention. JAMA PEDIATR. 2015;169(3):220–9. pmid:25561016
  94. 94. Letouzey M, Foix-L’Hélias L, Torchin H, Mitha A, Morgan AS, Zeitlin J, et al. Cause of preterm birth and late-onset sepsis in very preterm infants: the EPIPAGE-2 cohort study. PEDIATR RES. 2021. pmid:33627822
  95. 95. Wallstrom G, Anderson KS, Labaer J. Biomarker discovery for heterogeneous diseases. CANCER EPIDEM BIOMAR. 2013;22(5):747–55. pmid:23462916
  96. 96. Dhar V. Data science and prediction. Commun ACM. 2013;56(12):64–73.
  97. 97. Challis JR, Lockwood CJ, Myatt L, Norman JE, Strauss JF, Petraglia F. Inflammation and Pregnancy. Reproductive Sciences. 2009;16(2):206–15. pmid:19208789
  98. 98. Alpaydin E. Introduction to machine learning. Ieee Xplore d, Mit Press p, ebrary I, editors: Cambridge, Massachusetts: MIT Press; Third edition.; 2014.
    • 99. Della Rosa PA, Miglioli C, Caglioni M, Tiberio F, Mosser KHH, Vignotto E, et al. A hierarchical procedure to select intrauterine and extrauterine factors for methodological validation of preterm birth risk estimation. BMC Pregnancy Childbirth. 2021;21(1):306. pmid:33863296
    • 100. Stewart L, Moher D, Shekelle P. Why prospective registration of systematic reviews makes sense. Syst Rev. 2012;1:7. pmid:22588008
    • 101. Polettini J, Cobo T, Kacerovsky M, Vinturache AE, Laudanski P, Peelen MJCS, et al. Biomarkers of spontaneous preterm birth: A systematic review of studies using multiplex analysis. Journal of Perinatal Medicine. 2017;45(1):71–84. pmid:27514075
    • 102. Lucaroni F, Morciano L, Rizzo G, F DA, Buonuomo E, Palombi L, et al. Biomarkers for predicting spontaneous preterm birth: an umbrella systematic review. J Matern Fetal Neonatal Med. 2018;31(6):726–34. pmid:28274163
    • 103. Silagy CA, Middleton P, Hopewell S. Publishing protocols of systematic reviews: comparing what was done to what was planned. Jama. 2002;287(21):2831–4. pmid:12038926
    • 104. McAuley L, Pham B, Tugwell P, Moher D. Does the inclusion of grey literature influence estimates of intervention effectiveness reported in meta-analyses? Lancet. 2000;356(9237):1228–31. pmid:11072941