Background: About 15 million preterm births (PTB, < 259 days gestation), occur annually and it is the most notable cause of perinatal morbidity and mortality worldwide.¹ Changes in the vaginal microflora have been implicated in the pathogenesis of ascending intrauterine infection leading to preterm labour (PTL).² The metabolic activities of the vaginal microniche during pregnancy can provide insight into the pathogenesis of PTB as well as in identifying women at risk.³ These metabolic changes which are reflective of vaginal bacterial community function can be detected by Magnetic Resonance Spectroscopy (¹H-MRS).⁴

Aim: To assess the correlation between cervicovaginal fluid (CVF) metabolites and gestational age at delivery (GAAD) by ¹H-MRS.

Methods: A pair of high-vaginal swabs was obtained from pregnant women with intact membranes and no evidence of genital tract infection and grouped as follows:

Group 1 - women assessed at 20-22 gestational weeks (g.w.), n = 235; as part of their regular antenatal care;

Group 2 – a subset of group 1 identified as high-risk for PTB (due to previous history of PTB), assessed at 26-28 g.w., n = 91;

Group 3 - women presenting with threatened PTL, assessed at 24-36 g.w., n = 86.

CVF samples were dissolved in 600 µl phosphate buffered saline and a 400 µl MR sample containing 5% D₂O was analyzed with a 9.4T MR spectrometer, with 5mm observed probe using a Watergate water suppression pulse sequence (NS = 256, D1 = 5s, AQ = 1s, SW = 20.6 ppm, TD = 16446). ¹H-MR metabolites were integrated and normalized to the total spectrum integral (omitting the residual water signal) (Figure 1). Additional clinical parameters, e.g. cervical length, fetal fibronectin, gestational age at presentation (GAAP), were also recorded for each patient were consent was given. All identified integrals and clinical measurements were correlated against each other, using custom Matlab software, to produce a color map of Pearson correlation coefficient and significance (Figure 2).

Results: From the ¹H-MR spectrum of CVF, 8 metabolites (lactate, acetate, succinate, formate, glucose, glutamine/glutamate, alanine, and branched chain amino acids) were identified for analysis. As expected, fetal fibronectin and cervical length correlated with GAAD in all study cohorts, while vaginal pH correlated with GAAD among women in Groups 1 (20-22 g.w.) and 2 (26-28 g.w.) only (Table 1). Also, vaginal pH correlated with lactate and acetate integrals in all study cohorts, and with glucose, succinate/lactate ratio and glutamine/glutamate integrals in the Group 1 (20-22 g.w.) and Group 2 (26-28 g.w.) women only (Table 2). Interestingly, lactate (r = 0.15, p = 0.02) and glutamine/glutamate integrals (r = 0.14, p = 0.037) in the Group 1 (20-22 g.w.) women; and succinate/lactate ratio (r = -0.22 p = 0.03) in the Group 2 (26-28 g.w.) women, correlated modestly with GAAD. Furthermore, acetate vs. succinate integrals, lactate vs. glucose integrals, and alanine vs. branched chain amino acids, were all correlated across the groups (Table 3).

Conclusion: The interactions between the various metabolites and vaginal pH are indicative of the prevalent vaginal microbiota signature and the host-microbial interplay in the vaginal microecology. During pregnancy, inflammation of the fetal membranes (chorioamnionitis) due to female genital tract colonisation by organic acid producing anaerobic bacteria can lead to cervical remodelling, disruption of fetal membranes and leakage of fetal fibronectin. Hence, in addition to vaginal pH, fetal fibronectin, and cervical length, the anaerobic bacterial metabolic by-products particularly acetate, lactate, and succinate/lactate ratio could be useful biomarkers in identifying women at risk of delivering at earlier than normal gestational ages.