Background: Preterm birth (PTB) is a significant cause of neonatal death and disability worldwide. Sub-optimal vaginal microbiome profiles and inflammation are associated with increased spontaneous PTB risk, although the underlying mechanisms are not fully understood, and diagnostic tests are limited. This study aimed to identify host and microbial protein biomarkers in vaginal swab samples and characterise underlying mechanisms of spontaneous PTB in pregnant Australians.
Methods: Lateral mid-vaginal swabs were self-collected at mid-gestation by participants of the Preterm Birth Prevention Study (n=497). Luminex assays were used to measure levels of nine vaginal immune mediators (8 cytokines and matrix metalloproteinase-1), and K-means clustering was used to group women into “high” and “low” inflammation categories. TMTpro 18-plex liquid chromatography tandem mass spectrometry (LC-MS/MS) was conducted using a Q-Exactive Quadrupole-Orbitrap. Peptides were identified and proteins inferred using Proteome Discoverer and a custom database of human, bacterial and fungal taxa. Statistical comparisons were performed using Mann-Whitney U tests, Spearman’s rank correlations and limma, with p-values adjusted for multiple comparisons or corrected for false discovery rate (FDR).
Results: Overall, 7,452 host and 3,646 microbial proteins from 60 taxa were detected by LC-MS/MS. Multiple factors were associated with levels of vaginal inflammation mediators, including body mass index, gravidity, and maternal age, with the strongest correlates being metaproteomic host and microbiome profiles and reported recent sexual activity. When high and low inflammation groups were compared, 926 host and 420 microbial proteins were differentially abundant (FDR adj. p<0.05). These included overabundant host inflammatory proteins and bacterial proteins from non-optimal species, in women with high inflammation. In swabs collected within 24 hours of reported vaginal intercourse, 86 proteins were significantly differentially abundant, including prostate-specific antigen (FDR adj. p=2.20X10-23) and semenogellin 1 and 2 (FDR adj. p=5.83X10-29, p=4.80X10-30), plus a variety of host or male partner immune factors. None of the cytokines assessed or individual proteins identified were significantly associated with spontaneous PTB in univariate analyses; however, functional and multivariable analyses are ongoing.
Conclusion: There is an urgent need for new diagnostic approaches to better identify women at high risk of spontaneous PTB and enable earlier intervention. While we found multiple factors associated with vaginal inflammation in pregnancy, individually none exhibited prognostic value. Evaluation of microbial functional activity and incorporation of machine learning approaches may identify novel biomarker algorithms with future prognostic potential.