A project to validate the GLU test for preterm birth prediction in First Nations women

Keywords: Australian First Nations, diagnostic test, genotype, microbiome, pregnancy, preterm labour, preterm premature rupture of membranes, real-time polymerase chain reaction, vagina.

Preterm birth (PTB; delivery before 37 gestational weeks) is the leading cause of death and disability in children under 5 years of age. Globally, approximately 10% of infants are born too early.¹ Australian First Nations (the Aboriginal and Torres Strait Islander people of Australia) women are at far greater risk of this major complication of pregnancy. Nationally, in 2019, 13% of babies born to First Nations mothers were preterm, compared with 8.3% of babies from non-First Nations mothers.² The PTB incidence is even greater for First Nations women in regional and remote parts of the Northern Territory (NT) and Western Australia (WA) with national data reporting rates up to 23%³ (Fig. 1). The risk factors for PTB in First Nations women are multifactorial and not completely understood. Much of the discrepancy is likely the result of disparities in social determinants of health, fuelled by poverty, racism, and intergenerational trauma.⁴ However, at least one-quarter of all PTBs, but especially those that occur due to spontaneous preterm labour (sPTB), are attributed to intrauterine bacterial infection, especially at earlier gestational ages (GA).⁵

Fig. 1.  Preterm births among babies born to First Nations mothers by region 2016–2017. Source: AIHW.³

Numerous bacterial taxa have been implicated, with the bulk of these represented by Ureaplasma spp. and anaerobic organisms associated with a dysbiotic vaginal state.⁶ Conversely, dominance of certain Lactobacillus spp. in the vagina during pregnancy may offer some level of protection from sPTB.⁷ However, the vaginal bacterial microbiome varies substantially between ethnic cohorts,⁸ and little is known about its composition in First Nations women; the only study completed to date recruited 23 pregnant Australian First Nations women and reported a vaginal microbiome composition not dissimilar to that observed amongst African–American women, characterised by Lactobacillus spp.-depletion and presence of diverse anaerobic genera.⁹

Although infection may result in sPTB,¹⁰ it is the mother’s inflammatory response to infection that triggers preterm labour.¹¹ Whereas data exist on uterine inflammation during infectious and non-infectious scenarios,¹² very little is known about the pregnant vaginal environment relative to the microbial profile.¹³ Based on studies in non-pregnant women, McKinnon et al. have defined that a non-optimal vaginal microbiome elicits a pro-inflammatory milieu in the cervicovagina in contrast to an optimal vaginal microbiota that is non-inflammatory.¹⁴ Hence, an understanding of the vaginal pro- and anti-inflammatory state in pregnancy, alongside its microbial composition and function, may identify additional host and microbial biomarkers for sPTB prediction, as well as furthering our knowledge of the mechanistic nature of infection-related sPTB, which remains poorly understood.

Payne et al. recently conducted the largest mid-pregnancy vaginal microbiology study to date, the Predict1000 study, which resulted in the development of a novel mid-gestation vaginal microbial DNA test for prediction of Australian women at high sPTB risk, the GLU test (Fig. 2).¹⁵ In a cohort of 936 women, this test was able to detect women at risk of sPTB <37 weeks’ GA and ≤34 weeks’ GA, with sensitivities of 37.9 and 44.4%, respectively. More importantly, women who had a prior history of sPTB, our current best clinical predictor of sPTB risk, were almost as likely to deliver via sPTB as women who were GLU-positive in mid-gestation (adjusted odds ratio 3.61 vs 3.28). This test now underpins a National Health and Medical Research Council-funded randomised controlled trial (RCT) (ACTRN12617001593325) to assess the efficacy of a novel antimicrobial and probiotic treatment regimen in GLU-positive women to reduce sPTB risk. However, data on performance of the GLU test came from a predominantly Australian Caucasian cohort; only 22 women recruited to the Predict1000 study identified as First Nation Australians (2.3%).

Fig. 2.  The GLU algorithm.

Further research is needed to document the vaginal microbiome and associated inflammatory state in pregnant First Nations women. This includes an assessment of whether the GLU-test in its current form can identify First Nations women at increased risk of sPTB, and if not, ascertain if different host and microbial biomarkers can be identified that may be useful for this purpose.

Very few tools are currently available to identify women at high risk of sPTB during pregnancy. Identification of vaginal microbial biomarkers during mid-pregnancy has high potential for translation into clinically relevant tools that, importantly, can be acted on with suitable interventions to either prevent sPTB from occurring or delay the onset of preterm labour.

This research aims to develop approaches to identify pregnant First Nations women at risk of sPTB based on vaginal host and microbial biomarkers evaluated in early to mid-gestation. The specific research objectives are to:

1. Assess performance of the GLU test for prediction of sPTB in First Nations women.

2. Characterise the early to mid-gestation vaginal microbiome, inflammatory state, and host/microbial protein profile and document associations between these and maternal factors (such as smoking and diabetes).

3. Identify microbial and host biomarkers that will enable modification of the existing GLU test to enhance sPTB prediction in First Nations women.

To address the aims and objectives we will conduct a prospective cohort study.

In the Northern Territory, recruitment will occur from Royal Darwin Hospital (RDH), and Gove District Hospital (GDH). Both hospitals service towns and communities across the Top End. Approximately 600 First Nations mothers have babies across these two hospitals per annum.¹⁶

In Western Australia, recruitment will occur from the Geraldton Regional Aboriginal Medical Service (GRAMS), a community health service for First Nations peoples. GRAMS midwives provide antenatal care for over 100 pregnancies per year.

Pregnant First Nations women aged 16+ years attending antenatal clinics at ≤24 weeks’ GA will be invited to participate. Exclusion criteria include multiple pregnancies, current symptomatic vaginal infections, current or recent (preceding 2 weeks) antibiotic/antimycotic use, cervical sutures, high dependence on medical care, illicit drug use, and lack of capacity to provide written informed consent.

The anticipated recruitment number is 750 women, 500 at RDH/GDH and 250 at GRAMS. Assuming a baseline sPTB rate in Aboriginal women of ~7%, a cohort of 750 will attain ≥90% power to detect an increase in sPTB probability associated with a GLU (or modified version) positive test by an odds ratio ≥2.25 (increase from 7 to 14.5% sPTB risk), while simultaneously adjusting for other relevant microbial and clinical risk factors with partial r² = 0.1.

Participating women will first complete a medical/lifestyle questionnaire enquiring about key clinical and environmental factors that may impact the vaginal microbiome and pregnancy outcome. These include diabetes (type 1, type 2, or gestational), smoking and alcohol intake (assessed semi-quantitatively), and area of primary residence (urban vs rural). Women will then provide two self-collected vaginal swabs, one in liquid Amies media (COPAN e-Swab) for vaginal microbiome profiling and GLU analysis and one in QIAGEN Allprotect media for cytokine, chemokine, and host and microbial protein content analyses. A vaginal pH reading will also be obtained using a self-test kit (Canesten).

DNA will be extracted from swab 1 and GLU-status assessed via qPCR.¹⁵ The vaginal microbiome will then be characterised using full-length 16S rRNA gene sequencing on the PacBio Sequel II as per Goldenberg et al.¹⁰ Sequences will be analysed using Mothur (v1.47).¹⁷

Swab 2 (n = 180) will undergo cytokine and chemokine quantification and metaproteomic analyses to define the host inflammatory state and microbial functional properties. Samples will be selected based on pregnancy outcome, with a 60:120 split between sPTB (all cases in the cohort) and term births. This subgroup will attain ≥80% power to detect medium effect sizes for binary descriptors and ≥95% power to detect a difference of 0.5 s.d. for continuous descriptors between groups. Cytokines and chemokines will be analysed using a multiplex Luminex assay; targets include IL-1α, IL-1β, IL-6, IL-8, TNF-α, IL-1RA, RANTES, CXCL10, MCP-1, MIP-1α, MIP-1β and MIP-3α. Metaproteomic analyses will involve shotgun liquid chromatography tandem mass spectrometry using a Q-Exactive Quadrupole-Orbitrap mass spectrometer as per Alisoltani et al.¹⁸

The primary clinical end-point in this study is sPTB, defined as the spontaneous commencement of labour before 37 weeks’ gestation. Secondary end-points include established PTB research Core Outcome Measures,¹⁹ including sPTB ≤34- and 28-weeks GA. All end-points will be measured by obtaining delivery outcome data from NT and WA hospital pregnancy databases and patient medical records.

Evaluations of the effects of microorganisms, cytokines, host and microbial proteins, and maternal characteristics on the timing of birth will be conducted using linear, logistic and Cox proportional hazards regressions, as appropriate for gestational age at birth or sPTB. These regressions will be supplemented with recursive partitioning models, such as binary, regression and survival trees, designed to explore the non-linear relationships within the laboratory data alone and when combined with other obstetrics risk factors.

Ethical approval to conduct this study was granted by the Human Research Ethics Committee of the Northern Territory Department of Health and Menzies School of Health Research (HREC-2020-3659), and the Western Australian Aboriginal Health Ethics Committee (HREC-937). The NT study was approved by the Menzies School of Health Research Child Health First Nations Reference Group.

PTB has devastating impacts on First Nations families and communities, with significant long- and short-term complications. There are substantial emotional, psychological, and financial costs to families and communities. The cost to health services remains high. The continued poor rates are unacceptable. Further understanding of the pathophysiology is required. The presence and abundance of some bacterial pathogens in the mid-trimester vaginal microbiome has been shown to increase the risk of PTB in Caucasian women but this information is yet to be validated in First Nations women. Data generated from this study will confirm whether the GLU test in its current form is suitable for prediction of First Nations women at increased sPTB risk. In the case that the test is unsuitable for this cohort, additional data from 16S rRNA gene and inflammatory marker/protein profiling may allow cohort-specific microbial DNA and host and microbial protein signatures to be identified that predict First Nations women at high sPTB risk and who may benefit from mid-gestation treatment with specific antimicrobials/probiotics. Additionally, identifying protein biomarkers could be used to develop true point-of-care tests that are low cost and convenient to use in very remote settings where PTB rates are highest. This is likely to prolong pregnancy for many women and ultimately reduce mortality and morbidity for hundreds of infants each year.

Data sharing is not applicable as as no new data were generated or analysed during this study.

The authors declare no conflicts of interest.

This is a National Health and Medical Research Council (NHMRC) funded project.